X-Planes: X-67, the Slot Taken by XQ-67A

The whole public case for the X-67 having been skipped is one sentence, and four fifths of that sentence is a comparison with another number. The entry by the compiler, Andreas Parsch, reads that, just like the X-58, the X-67 slot was skipped after the allocation of the XQ-67A to the General Atomics Off-Board Sensing Station unmanned aircraft [Missing USAF and DOD Aircraft Designations]. The entry ends there. There is no reasoning about this case, no statement of confidence, and no claim that the aeroplane does not exist.

Set that beside the entry one paragraph above it. The X-58’s entry states that the designation has been skipped, names the aeroplane and the laboratory that took the number, and adds that it is practically certain that the number was chosen to fit into the research series even though a vehicle designated with the unmanned symbol should have taken an unmanned number [Missing USAF and DOD Aircraft Designations]. Reasoning, a confidence grading and a claim about emptiness, and the X-67’s entry carries none of the three.

The comparison is the point and a uniqueness claim is not available. The compiler reasons by analogy elsewhere. His note on the XRQ-73A derives the number 73 from the XRQ-72A because one programme follows the other, and his note on the YFQ-44A says that the number is out of sequence like the YFQ-42A’s and that why 43 was skipped is unclear [Parsch, DOD 4120.15-L Addendum]. What is distinctive here is narrower. The 73 is derived from a programme relationship, so the reasoning concerns its own case. The 67 is derived from another number’s derivation, and that other derivation is the only place the reasoning lives.

This is the sixty-eighth article in the X-Planes series, following the X-1, the X-2, the X-3, the X-4, the X-5, the X-6, the X-7, the X-8, the X-9, the X-10, the X-11, the X-12, the X-13, the X-14, the X-15, the X-16, the X-17, the X-18, the X-19, the X-20, the X-21, the X-22, the X-23, the X-24, the X-25, the X-26, the X-27, the X-28, the X-29, the X-30, the X-31, the X-32, the X-33, the X-34, the X-35, the X-36, the X-37, the X-38, the X-39, the X-40, the X-41, the X-42, the X-43, the X-44, the X-45, the X-46, the X-47, the X-48, the X-49, the X-50, the X-51, the X-52, the X-53, the X-54, the X-55, the X-56, the X-57, the X-58, the X-59, the X-60, the X-61, the X-62, the X-63, the X-64, the X-65, and the X-66.

The X-39, the X-52 and the X-58 each reached this series as a number rather than as an aeroplane, and the X-67 is the fourth. The first was reserved and lost because nobody wrote the letter that would have assigned it. The second was requested in writing and refused for resembling a number already in use. The third was skipped while the number went to an unmanned aircraft in a neighbouring series. The fourth is the third happening again, and a second occurrence matters because it is the point at which a precedent becomes a rule.

This article is therefore not a second account of the same mechanism. The X-58 article established the mechanism, measured the family of borrowed numbers, and showed that scarcity explains nothing. What it could not do, having only one case, is ask whether the case was an accident. The X-67 makes that question answerable, and the answer turns out to need three things the earlier article did not have.

The first is a test. The claim that a number was taken from the research series is checkable against the register, because a borrowed number should coincide with the number the research series was about to issue and should coincide with nothing else. Applied to every out-of-sequence number in the unmanned series, that test fires two times in six, and both times it names the research series. The compiler asserts the derivation for the X-67 and offers no evidence. The register he compiled supplies the evidence he does not cite.

The second is the instruction. The governing publication defines the next number to request from the last approved design number of the same basic mission, and it refuses requests in skipped sequences [DAFI 16-401]. Read together, those two sentences mean that nobody had to decide to skip the X-67. What was needed was for somebody to ask for the X-68, after which the X-67 became unrequestable for ever. The skip is a consequence of a later request rather than an earlier decision, and the article that said the absence was permitted and unexplained can now say what permitted it and when it became irreversible.

The third is the aeroplane, and only for one reason. The XQ-67A flew on 2024-02-28 and it was built to prove that several different aircraft can be raised on one shared core structure, an idea its programme called a genus carrying species [AFRL first flight release] [GA-ASI first flight release]. A design number exists to answer the question of when a thing is a new thing. The aeroplane that consumed this one was built to prove that the answer is modular. That collision is this article’s subject, and it gives an anomaly with four sentences of record an argument.

No X-67 vehicle was ever built, so there is no keystone to identify and no system to dimension, and the sections on sizing, dependent systems and flight say so and stop. The aeroplane that took the number is treated as evidence about the number and not as the subject.

The Research Question

The X-67 is a number and a number has no research question, so this section has to separate two things the title of the article joins. No vehicle was ever designated X-67, and the register carries no research-series row with the design number 67 [Parsch, DOD 4120.15-L Addendum]. Nothing was built to answer a question under that number, and the record attributes no question to it.

The aeroplane that took the number did have a question, and the laboratory stated it. The XQ-67A existed to show that several aircraft can be raised as species on one shared genus chassis, and the laboratory’s release says that the object was the method as much as the aeroplane [AFRL first flight release]. That question is the XQ-67A’s and not the X-67’s.

The question this article asks is therefore about the number, and the opening states it. Whether a second borrowing from the research series was an accident, what permitted the skip and when it became irreversible, and what a design number indexes at all are the questions taken up under Why the Slot Is Empty, So Far As the Record Says and What the Data Changed. The aeroplane’s question enters only because its answer bears on the last of them.

Programme Origin, What the Record Says Happened

The public record of the X-67 consists of one absence and four sentences. The absence is in the allocation register and the sentences are in two compilations by the same author, so the argument that follows has a narrow base and the article states this at the outset.

The Register Is the Primary Source and the Primary Source Is a Table

The register is a private compilation by Andreas Parsch that continues DOD 4120.15-L, the Department of Defense list of model designations, and it covers allocations from 19 August 1998 onward [Parsch, DOD 4120.15-L Addendum]. Its descriptions of later allocations are marked by the compiler as not official Department wording, as described below, so the register is evidence about the Department’s allocations and not a Department publication. Its front matter explains why it exists. The Department published no updated edition for almost six years after October 1998, the compiler requested the new assignments under the Freedom of Information Act, and the list has been maintained since. The compiler also records that the Department announced in 2018 that the printed list would be retired and that the data would be made available through other online sources, and that those sources have not been implemented [Parsch, DOD 4120.15-L Addendum]. That last sentence turns out to matter and the article returns to it.

Every register count in this article is of the register as last updated on 23 August 2026, a date after this article’s. The page as archived on 10 December 2025, two days before that date, carries 526 designation rows, 29 of them in the research series and the last of those the X-66A [Parsch, DOD 4120.15-L Addendum, as archived on 10 December 2025], and the dated note under The Article’s Own Vantage Point records when the newer research rows appeared. Five rows of the later copy carry a partial date in 2026, after this article’s date, and they enter its row counts but no chronological statement.

The table carries 539 designation rows, of which 532 have a designation in the standard form and seven do not, the latter being the rows listed below exactly as the register prints them.

AGM-158B-2
AGM-158B-3
AIM-120ER
AMH-130R AMH-6R
C-130J(CC-130J)
F-15EX
W-2A WS-2(?)A

Of the well-formed rows 503 carry a full day, month and year and 29 carry a partial date or none. The dated rows run from 1997-12-10 to 2025-11-21. Every chronological statement below uses the dated subset and every count uses the whole, because a row with a partial date is still an allocation and excluding it would understate the register while including it would corrupt an ordering.

The register also marks its reliability, in four states, and this article needs all four. The compiler’s note explains that no complete official data is available for designations assigned after October 2018, that for later allocations the official description is no longer releasable to the public, and that such descriptions are therefore shown in blue to indicate that they are not official Department wording [Parsch, DOD 4120.15-L Addendum]. Rows missing from officially released data altogether are shown entirely in blue, a stronger claim because it concerns a whole allocation rather than one cell.

The four states are exhaustive and disjoint, and any officiality count depends on that, so the article states the partition outright.

So 103 of the 539 rows carry a mark of some kind, and five of them are among the seven rows in non-standard form, which the table counts with the rest. The distinction between the second state and the fourth is the one this article leans on, and merging them would destroy the comparison drawn under The Second Case Falls on the Far Side of a Line the First Did Not. The four counts in the table sum to the 539 rows, so a row falling into none of the four or into two of them would show as a total that no longer matches.

The Absence Is Measured and Not Quoted

An absence cannot be quoted, so it has to be measured. Searching the register’s text for the string X-67 finds nothing at all, not even a denial. The two occurrences of that string anywhere in the compiler’s work are both in the separate compilation of missing designations [Parsch, DOD 4120.15-L Addendum] [Missing USAF and DOD Aircraft Designations]. And a search would prove nothing about the table in any case, because a note is not a row. The measurement is an enumeration of every row whose basic mission symbol is the research symbol.

There are 31 such rows and they carry 25 distinct design numbers. The numbers run from 37 to 76, which is a span of 40 positions, and fifteen of those positions carry no row at all, being 38, 39, 41, 42, 43, 52, 58, 67, 69, 70, 71, 72, 73, 74 and 75. The number 67 is among them, and the count of research-series rows carrying the design number 67 is zero.

That absence divides into two kinds and only one kind is evidence, and the division is a disjoint decomposition of the set above.

The register opens on 19 August 1998, so a designation allocated before that date is absent from it for a reason that has nothing to do with being skipped. Five of the fifteen absences fall in that category, being 38, 39, 41, 42 and 43. Ten fall inside the window, being 52, 58, 67, 69, 70, 71, 72, 73, 74 and 75, and for those the absence is informative because their neighbours are in the same document.

That division is a documented fact, because the government published its own list and the list carries them. DoD 4120.15-L of 12 May 2004 lists the X-41A as an experimental manoeuvring reentry vehicle, the X-42A as an experimental expendable vehicle and the X-43A to Microcraft as a hypersonic experimental research aircraft, each with a sponsor and an engine entry [DoD 4120.15-L, 2004 with Change 1]. So three of the five absences outside the window are allocations the Department itself printed, and their absence from a register that opens in August 1998 dates them and casts no doubt on them.

The share deserves a figure because it is larger than a reader would guess. It is the ratio of the two counts above. Writing the set of absent numbers as $A$ and the span’s endpoints as $d_{\min}$ and $d_{\max}$ gives the completeness of the record directly.

That evaluates to 37.5 percent. More than a third of the positions in the span the register covers carry no research-series row at all.

The series is not fast, so the share matters. The allocation rate over the window is the count of dated rows divided by the interval they span.

The 29 dated research rows span 26.289 years, so $\lambda$ is 1.1031 allocations per year. Dividing the numbers lost inside the window by that rate converts them into years of the series’ output.

That is 9.07 years. A series issuing roughly one number a year has lost ten numbers inside the register’s window, a decade of its output.

That answers a different question from the arithmetic of the X-58 article, which computed the rate at which the series consumes numbers and the headroom before the design-number field must widen, and concluded that scarcity explains nothing. The missing share says nothing about scarcity and something about the record.

The X-67’s absence is therefore dispositive in a way that the X-41’s is not. The X-66A was allocated in May 2023 and the X-68A in August 2025, both inside the window and both in the same table. A row between them would have been recorded if it existed. That is the strongest statement the register supports about the X-67, and it concerns the register rather than an aeroplane.

The Number Was Taken While It Was Still the Next One Due

The allocation that carries the number is in the unmanned series and the register gives its date. The XQ-67A was allocated on 2023-07-19 to General Atomics, with a sponsor cell naming the United States Air Force, printed as USAF, an engine cell reading 1 Williams FJ44-4A, and a description reading that it is an unmanned air vehicle for the Off-Board Sensing Station programme [Parsch, DOD 4120.15-L Addendum].

The research series had reached 66 at that moment. The X-66A was allocated on 2023-05-03, 77 days before the borrowing, and no research allocation after that one precedes it. So the next research design number due on 2023-07-19 was 67, and 67 is the number the unmanned allocation took.

The same arithmetic holds for the earlier case and the two differ in their timing. The XQ-58A was allocated on 2017-07-10 to Composite Engineering under the same sponsor, the research series had reached 57 through the X-57A of 2016-05-03, and the next number due was 58 [Parsch, DOD 4120.15-L Addendum]. The gap between the preceding research allocation and the borrowing was 433 days in the first case and 77 days in the second, a ratio of 5.623. The second number was taken while it was much fresher.

And the research series then waited. After the first borrowing the next research allocation was the X-59A on 2018-06-21, 346 days later. After the second it was the X-68A on 2025-08-20, 763 days later, the longer of the two and the interval during which nothing in the record had yet treated the X-67 as spent.

Two Designations in the Whole Register Carry That Prefix and Nothing Else

The pair is small and bounded and the register says so by enumeration. A designation is a run of letters, a design number and a series letter, and the last letter before the dash is the basic mission symbol while every letter ahead of it is a prefix [DAFI 16-401]. Searching the register for rows whose basic mission is the unmanned symbol and whose prefix is exactly the experimental symbol returns two designations, being XQ-58A and XQ-67A.

That matters because the register carries other rows combining the two letters and they are not the same thing. The XRQ-22A, the XRQ-72A and the XRQ-73A carry a reconnaissance symbol between the two, and the XMQ-17A and the XMQ-19A carry a multi-mission symbol. The two bare forms are the only designations in which the experimental prefix sits directly against the unmanned mission with nothing between, and they are the two numbers the compiler says came from the research series.

The Register States the Derivation Twice and the Second Statement Is Thinner

The compiler writes about the X-67 in two places and neither sentence contains an argument. In the register’s end notes the entry reads that the out-of-sequence number 67 of the XQ-67A was taken from the X-series [Parsch, DOD 4120.15-L Addendum]. In the compilation of missing designations the entry reads that, just like the X-58, the X-67 slot was skipped after the allocation of the XQ-67A to the General Atomics Off-Board Sensing Station unmanned aircraft [Missing USAF and DOD Aircraft Designations].

Set those beside what the same compiler wrote about the first case. The note against the XQ-58A reads that the out-of-sequence number 58 of the Kratos Valkyrie was taken from the X-series, and it adds a parenthesis stating that there is no X-58 [Parsch, DOD 4120.15-L Addendum]. The missing-designations entry reads that the X-58 designation has been skipped, that the designation XQ-58A was assigned to the Valkyrie unmanned aircraft built for the Air Force Research Laboratory, and that it is practically certain that the number 58 was chosen to fit into the X-series even though a vehicle designated with the unmanned symbol should get a number in the unmanned series [Missing USAF and DOD Aircraft Designations].

Three things are present in the first case and absent in the second. The reasoning, that an unmanned designation should have taken an unmanned number. The confidence, stated as practically certain. And the claim that the slot is empty, which the first note makes as a parenthesis and the second does not make at all. What the second case has instead is the phrase just like the X-58.

The compiler is not being careless, and his treatment of the rest of the family shows it. He uses a graded vocabulary and he uses it consistently.

He writes apparently, probably, practically certain, unclear and nothing at all, and he writes each where the evidence calls for it [Parsch, DOD 4120.15-L Addendum] [Missing USAF and DOD Aircraft Designations]. Because the XRQ-72A’s entry states plainly that the derivation is unknown, the silence against the X-67 is informative. A compiler who will write unclear when he cannot derive a number is not withholding an argument. He is citing one.

And the citation runs one step deep rather than two. The 73 is derived from the 72, whose origin is recorded as unknown, so that derivation stands on a number and not on a reason. The 67 is derived from the 58, whose origin is recorded with reasoning and a confidence. Of the two chains in the family, the chain this article is about has the sound root.

The Second Case Falls on the Far Side of a Line the First Did Not

This difference between the two borrowings matters most to a register, and the notes do not mention it. The XQ-58A was allocated in July 2017 and its description is official Department wording. It reads that the aeroplane is an unmanned, transonic, long-range, low-cost, limited life, strike unmanned aircraft system, and it carries the popular name Valkyrie [Parsch, DOD 4120.15-L Addendum]. The XQ-67A was allocated in July 2023 and its description carries the mark saying it is not official Department wording.

In the first case the government said what the aeroplane was for and in the second it did not. The register’s state for the XQ-58A’s description is official and for the XQ-67A’s it is cell. The compiler’s phrase just like the X-58 therefore draws an analogy across the evidentiary boundary his own front matter declares.

The research series crosses the same boundary in the same place and this series has measured it. Of the 31 research rows, 21 are fully official, seven are official allocations whose descriptions are the compiler’s, one is official with a partly unofficial description, and two are rows absent from officially released data altogether. The last research row with an official description is the X-60A of September 2018 and the first without one is the X-61A of June 2019. The X-61 article recorded that finding, and this article confirms it by recomputation.

This measurement also refines a figure other articles in this series state, being the research series’ officiality split as twenty-one official, nine wholly unofficial and one partly. The register gives 21 official, nine wholly unofficial and one partly, which agrees exactly, provided that wholly unofficial merges two different claims. Two of the nine are rows absent from officially released data altogether and seven are official allocations whose stated purpose is the compiler’s. The recorded figure is right and the category is too coarse for this article’s use, because the XQ-67A falls in the second group and the comparison above consists of that distinction.

Some of the row survives the mark. The date, the designation, the contractor and the engine data on the XQ-67A’s row are unmarked, so they are official. The register officially records that the aeroplane has one Williams FJ44-4A and unofficially records what it is for. That inversion is the register’s shape after 2018, and it explains why an article about a designation can still be written while an article about a purpose cannot.

The Aeroplane That Took the Number, in the Record’s Own Words

The airframe enters this article as evidence about what a design number is for, because the programme that consumed this number existed to show that a design can be a shared core with interchangeable mission kits.

The laboratory’s release is the primary account and it is short. The Air Force Research Laboratory’s Aerospace Systems Directorate flew the XQ-67A, an Off-Board Sensing Station unmanned air vehicle, on 2024-02-28 at the General Atomics Gray Butte Flight Operations Facility near Palmdale, California [AFRL first flight release]. The aircraft completed several test points and recovered safely on the first of a series of flight tests.

The release then states the claim this article is about. The XQ-67A is described as the first of a second generation of autonomous collaborative platforms, and the laboratory’s capability lead is quoted saying that, following the success of the XQ-58A, the XQ-67A proves the common chassis or genus approach to aircraft design, build and test, and that this approach paves the way for other aircraft species to be rapidly replicated on a standard genus chassis [AFRL first flight release]. The same official is quoted explaining the mechanism in terms borrowed from another industry.

This approach will help save time and money by leveraging standard substructures and subsystems, similar to how the automotive industry builds a product line. From there, the genus can be built upon for other aircraft, similar to that of a vehicle frame, with the possibility of adding different aircraft kits to the frame, such as an Off-Board Sensing Station or Off-Board Weapon Station.

The contractor’s release says the same thing in the language of a first instance. General Atomics Aeronautical Systems states that it flew the aeroplane on 2024-02-28, that the Off-Board Sensing Station is a laboratory programme and that the firm was selected in 2021 to design, build and fly the aircraft, and that with the flight it has validated the genus and species concept first developed with the laboratory under the Low-Cost Attritable Aircraft Platform Sharing programme [GA-ASI first flight release]. The programme manager is quoted saying that the aeroplane is the first species to be designed and built from this shared platform, and the contractor’s vice-president of advanced programmes that it is the first aircraft type built and flown using a common core chassis that promotes commonality across multiple vehicle types.

The lineage the laboratory gives runs through four programme names and the article uses it. The Low Cost Attritable Aircraft Technologies initiative began with in-house designs in late 2014 and early 2015, and its early studies evolved into the requirements definition for the Low Cost Attritable Strike Demonstrator joint capability technology demonstration, which produced the XQ-58 for its first flight in 2019. The platform-sharing programme began as discussions in 2018 and fed into the Off-Board Sensing Station programme, which culminated in building and flying the XQ-67A [AFRL first flight release]. The laboratory also records that the team initially worked with five industry vendors and that it decided at the end of 2021 to exercise the opportunity to build the General Atomics design.

One sentence in that release states the programme’s object more plainly than any other, and the object is the method as much as the aeroplane. The elision covers a clause allowing that there is an aircraft at the end of the programme which the laboratory expects to get a great deal of use from, and it does not change the sense [AFRL first flight release].

This has to be done affordably and this program … the purpose of this program was the journey of rapid, low-cost production as much as it was the destination of a relevant combat aircraft.

The aeroplane has flown again since and the second flight is about mission systems rather than about the airframe. In July 2025 the contractor reported that the aeroplane demonstrated the integration of government reference autonomy during a flight test in the California high desert, paired with active tactical datalink communications, and that it executed test points validating autonomy, mission computing, networking, power and thermal management and datalinks [GA-ASI autonomy release]. The release describes the aeroplane as a testbed and as critical to exploring the platform sharing approach, repeating the claim of the first release, this time for mission systems.

What the Award Record Adds That the Releases Do Not

The federal award record names the five vendors the laboratory’s release leaves anonymous, and it prices the down-select the releases describe without numbers. Querying it for the platform-sharing programme returns five contracts, all awarded in the spring of 2019 [USAspending].

The count of five matches the laboratory’s sentence about five industry vendors, and this article has found no other corroboration for it. The total is 24,479,631.19 dollar. Every share quoted in this section is the same relation, one award over the sum of them, written once here and not again.

The firm that eventually built the aeroplane already held 43.60 percent of that money, and the top two awards together held 83.53 percent. The eventual winner’s award was 1.0922 times the second largest and 7.710 times the fourth.

All five of those firms hold research-series designations in their own right, and between them they hold 23 of the register’s 31 research rows [Parsch, DOD 4120.15-L Addendum].

The smallest of the five platform-sharing awards went to Aurora Flight Sciences in March 2019 and that firm was allocated the X-65 four years later [USAspending] [Parsch, DOD 4120.15-L Addendum]. A laboratory programme about sharing airframe cores engaged, in one go, the firms that own three quarters of the experimental designation series.

Then the Off-Board Sensing Station contracts, and these are the ones that record the decision. The award record returns two, one day apart [USAspending].

The ratio between them is 4.2483, and the two figures deviate from what was reported at award in opposite directions. Trade coverage of the October 2021 awards reported matching cost-plus-fixed-fee contracts of 17,700,000.00 dollar each, potentially worth 49,000,000.00 dollar if an option of 31,400,000.00 dollar were exercised, which sums to 49,100,000.00 dollar, with a base effort to be delivered within a year and an option effort to be completed by the end of January 2024 [Kratos, General Atomics Get Contracts for Off-Board Sensing Station].

Comparing an outturn against a reported figure is the other relation this section uses repeatedly, a relative deviation from a reference.

Against those figures the award record shows one contract at 3.8410 times its reported base and 38.75 percent above its reported ceiling, an excess of 18,986,112.00 dollar, while the other ended at 90.41 percent of its reported base, 9.59 percent short. The two deviations have opposite signs, which is the shape of a competition resolved rather than completed and is the only thing this pair of numbers establishes. One contract grew past the maximum its option defined and the other stopped before its minimum. The sum of the two is 83,989,414.00 dollar, 3.431 times the total of the platform-sharing programme that preceded it.

The same coverage names the solicitation behind the awards. The awards followed a Broad Agency Announcement of September 2020 under a laboratory heading called Science and Technology for Autonomous Teammates, proposals were submitted in April 2021, and two awards were expected [Kratos, General Atomics Get Contracts for Off-Board Sensing Station]. That announcement is a primary document and has not been read.

And the Laboratory’s Own Account of When It Decided Disagrees With the Contemporary Record

The laboratory’s 2024 release states that it decided at the end of 2021 to exercise the opportunity to build the General Atomics design [AFRL first flight release]. Contemporary reporting places the selection in February 2023, describing the laboratory as having asked General Atomics to build and flight test its aircraft, and noting that the two development contracts of October 2021 had a twelve-month development period [AFRL Chooses General Atomics Drone for Off-Board Sensing Station Program].

Those two accounts are about a year apart and the article does not resolve them. A twelve-month base effort beginning in late October 2021 ends in late October 2022, so an option exercised after it could not have been exercised at the end of 2021 in any ordinary sense. The most likely readings are that the release’s end of 2021 refers to the award itself rather than to the option, or that it is an error for the end of 2022. The record contains no document that chooses, and the designation date of 2023-07-19 sits after both candidate dates and therefore distinguishes neither.

The Designation Arrived Three Quarters of the Way Through

Laying the award record beside the designation date and the first flight gives the ordering.

The designation came 631 days after the build contract and 224 days before first flight. Placing an event inside an interval takes a ratio of two differences, the last relation this section needs.

Measured against the 855 days from build contract to first flight, the number arrived with 73.8 percent of that interval already elapsed. The aeroplane was contracted, designed and largely built before it had a designation.

That is a sixth ordering between a number and an aeroplane. This series has recorded five already.

The sixth is the number arriving mid-build, after the design is committed and before it flies. It is probably the ordinary case for a modern demonstrator rather than an anomaly, and it is recorded because an ordinary case is the baseline the five unusual ones are unusual against, and a list of exceptions does not supply that baseline by itself.

The award record is also silent about the designation itself, as it was for the X-62. Querying it for the string XQ-67A returns no awards at all [USAspending], a result that article explains as a designation being a claim one part of a government makes to another, of no use to the part that pays the invoices.

Why the Slot Is Empty, So Far As the Record Says

The register says that the slot was skipped and does not say who skipped it. The governing instruction explains how such an absence is possible, and it explains it twice in two different mechanisms, and the article’s finding is that the second mechanism makes the first unnecessary.

The Instruction Defines the Next Number From the Last Approved One

The publication in force is Department of the Air Force Instruction 16-401 of 3 November 2020 [DAFI 16-401]. Its request procedure tells the requester what to look up before asking for anything.

Research and determine the “next-in-series” MDS designator to be requested … For standard aircraft, reference the basic mission designator and research the last approved design number and design series of that basic mission designator to determine the “next-in-series” MDS designator.

The elision covers a parenthetical naming where to look, a page on the Defense Technical Information Center’s DoDTechipedia site or the allocating office itself, and the article returns to that parenthetical below because the site it names is the half of the sentence that has not survived.

So the next number is defined from the last approved design number and not from the lowest unused one. A gap left in a basic mission is therefore not a slot waiting to be filled. It is behind the pointer, and the procedure never looks behind the pointer.

That is a stronger statement than the one this series has been making. The X-39 article found that reuse of a reserved designation requires cancellation first and that the cancellation was never filed, a finding that concerns a single paperwork failure. The procedural sentence above is general, and it implies that in the ordinary case there is nothing to cancel, because a skipped number was never approved and so can never be the last approved one.

And the Instruction Refuses to Sell a Skipped Number Back

One page earlier the eligibility criteria close the door explicitly.

The request pertains to a new designation or modification/revision to an existing designator. Requests are not accepted for designators in reverse or skipped sequences.

That sentence alone decides whether the X-67 could still be issued, and it decides against. Once the X-68A was approved, a request for the X-67 would be a request in a skipped sequence, which the instruction declines to accept. The number is not vacant. It is unrequestable.

One of those two sentences was cited by the X-58 article and the other was not, and the difference matters more than it looks. The X-58 article cited the research step, in its non-standard-aircraft branch, and used it to measure how far above the unmanned ceiling the borrowed number landed [DAFI 16-401]. It also cited the discretion clause, and concluded that the absence was permitted and unexplained rather than decided. What it did not cite is the eligibility criterion quoted above, and the irreversibility comes from that sentence. A pointer that only moves forward leaves a gap behind it. A rule refusing requests in skipped sequences is what makes the gap permanent, so the absence is not merely permitted but irreversible by rule.

The Rule Changed, and the Register Holds the Before-Case in the Same Series

The research series contains one number that was skipped and later recovered, and the register records both dates. The pointer advanced from 48 to 50 on 2002-02-13, passing the 49, and the X-49A was allocated on 2003-05-23, 464 days later [Parsch, DOD 4120.15-L Addendum]. The X-49 article records why the number was passed, namely that the round fifty was wanted for a rotorcraft, and that the vacancy was filled when a programme changed sponsor. What that article did not ask is whether the fill was permitted.

It was, and the edition then in force confirms it in one sentence. The joint instruction of 1 September 1997, which superseded the issue of 9 September 1994 and like it implements a Department directive of 1985, directs the coordinating office to assign and reserve the next available consecutive design number within each basic mission [AFJI 16-401, 1 September 1997]. The 1994 issue states the same procedure plainly [AFJI 16-401, 9 September 1994].

HQ ASC/ENOS will assign the next available consecutive design number within each basic mission for new vehicles.

Next available is not the same quantity as last approved, and this article’s mechanism lies in the difference. A number that has been passed is available, so under the rule of 1994 and 1997 the 49 was the next available number in May 2003 and the X-49A is the instruction obeyed rather than bent. Under the rule of 2014 and 2020 the next number is derived from the last approved one, and a passed number is behind the pointer and unrequestable.

The word skip appears in none of the issues of 1994, 1997 and 2005, as the X-52 article found when it read them for a refusal made in 2006 under the last of them [AFJI 16-401, 9 September 1994] [AFJI 16-401, 1 September 1997] [AFI 16-401(I), 14 April 2005]. That article dates the arrival of a bare discretion to skip to the issue of 16 May 2014, which also states that skipped design numbers cannot be requested or used [AFI 16-401, 16 May 2014]. This one adds what the same change did in the other direction. It closed the route the X-49A had used.

So the four research-series numbers passed before the X-76A fall under two regimes, the first two under the rule of the next available number and the last two under the rule that refuses skipped sequences, and only the last two were unrecoverable from the start.

The X-67 is therefore the second number in the research series to be skipped under a rule that makes skipping permanent, and the X-58 was the first. That dated claim from primary text is the only statement this article makes about why the slot is empty that carries both a date and a document. Neither the X-58’s slot nor the X-67’s was ever recoverable.

Six Editions Place the Change in 2014, and the Answer Narrows the Claim

Whether an edition between 1994 and 2020 changed the language earlier can be settled from the editions themselves. Six editions of the rule have been read and they divide into two groups.

The 2005 and 2014 editions settle it. Air Force Instruction 16-401(I) of 14 April 2005 states that the coordinating office will assign and reserve the next available consecutive design number within each basic mission for new vehicles, the 1997 wording and the 1994 wording with one verb added [AFI 16-401(I), 14 April 2005]. The word skip appears nowhere in it, and neither does any eligibility criterion refusing a request in a skipped sequence. The issue of 16 May 2014 supersedes it directly and carries both. It tells the requester to research the last approved design number to determine the next designator to request, it refuses requests in skipped sequences for vehicles with an extensive history of designations, and it reserves to the office the authority to skip a design number at discretion while stating that skipped design numbers cannot be requested or used [AFI 16-401, 16 May 2014].

The change was made by the issue of 16 May 2014, and it made two changes rather than one. The quantity moved from the next available number to one above the last approved number. And the duty moved from the agency to the requester. The X-49A of 2003 falls under the old form, and the X-58’s skip of 2018 and the X-67’s skip of 2025 both fall under the new one.

The 2020 issue keeps the refusal and words it more broadly. Its eligibility criterion refuses requests in reverse or skipped sequences without the 2014 limitation to vehicles with an extensive history of designations, and its skip clause keeps the reserved authority but drops the sentence saying that skipped design numbers cannot be requested or used [AFI 16-401, 16 May 2014] [DAFI 16-401]. Under either issue a passed research number is unrequestable, under the 2014 issue by both sentences and under the 2020 issue by the eligibility criterion alone.

The rule changed in the instruction, and the directive above it is silent on numbering. The 1994, 1997 and 2005 issues implement the Department directive of 2 May 1985, and the 2014 issue implements its reissue of 29 November 2004, which it cites as DoD Directive 4120.15E [AFJI 16-401, 9 September 1994] [AFJI 16-401, 1 September 1997] [AFI 16-401(I), 14 April 2005] [AFI 16-401, 16 May 2014]. The reissued directive, read as amended by Change 2 of 17 September 2020, requires a unique mission-design series designator for each vehicle and directs the Secretary of the Air Force to publish a joint regulation and a single Department-wide list of approved designators, and it says nothing about how a design number is chosen or whether one may be skipped [DoD Directive 4120.15, 29 November 2004]. The numbering rule is therefore set by the instruction alone.

The 2005 edition also carries the design-number example in a form the later editions changed by one word. It reads that the design number 16 in F-16A is the sixteenth MDS requested for an aircraft with a fighter mission under the current reporting system, implemented September 1962 [AFI 16-401(I), 14 April 2005]. The 2014 and 2020 editions say approved rather than requested [AFI 16-401, 16 May 2014] [DAFI 16-401]. A requested ordinal and an approved ordinal are different counts whenever a request is refused, and this series has written about a refusal.

One narrow caveat survives. The 1994 and 2005 sentences are about what the coordinating office will assign rather than about what a requester may ask for, so reading them as permitting a fill is an inference from a procedural duty. The X-49A is the evidence that the inference is right, because the fill happened.

Three Readings of the X-68A, and the Record Chooses None

The X-68A is the strongest evidence in the record that somebody treated the X-67 as spent, and better evidence than anything the compiler cites. It was allocated on 2025-08-20 to General Atomics with the Defense Advanced Research Projects Agency as sponsor [Parsch, DOD 4120.15-L Addendum]. At that date the last approved design number in the research series was still 66, because no X-67 row exists. Under the procedural sentence quoted above, the next-in-series designator was therefore X-67A, and the government approved X-68A.

There are exactly three readings of that and the record supports none of them over the others.

The first is that the design-number pool is shared across basic missions in practice. Under this reading the last approved design number relevant to the research series was 67, because the XQ-67A carries 67, and the next-in-series was correctly 68. This reading makes the borrowing real and makes the skip automatic. It is also the reading the instruction’s definition denies, since the instruction says the design number identifies major design changes within the same basic mission and illustrates it by saying that the 16 in F-16A represents the sixteenth approved designator for an aircraft with a fighter mission [DAFI 16-401].

The second is that the allocating office exercised its discretion to skip. The instruction reserves that authority without criterion, without a requirement to record a reason and without any appeal [DAFI 16-401]. Under this reading nothing consumed the X-67 and the X-68A’s number is a decision nobody has to explain. The record contains no exercise of the authority and the authority requires none to be recorded, so this reading predicts the absence of evidence found here.

The third is that the research-series pointer was read off a source that lists the XQ-67A. The procedure tells the requester to reference a page on a Department technical information site, or to contact the allocating office [DAFI 16-401]. A list sorted by design number within a vehicle type places the XQ-67A among the sixty-sevens, and the register this article uses sorts that way, by primary mission letter and then design number and then series letter [Parsch, DOD 4120.15-L Addendum]. Under this reading the skip is a consequence of how the list is ordered rather than of anybody’s intent.

The three readings differ in what they say about agency and agree in every observable. All three predict an X-68A, no X-67 row and no public document. The article therefore states that the X-67 is unrequestable, states when it became so, and declines to say who made it so.

The Register’s Own Next-Number Figure Cannot Settle It

It is tempting to settle the question with the compiler’s published statement, and the statement is circular for this purpose. He writes that the next available design number in the research series is X-69 [Missing USAF and DOD Aircraft Designations]. That figure presupposes that 67 and 68 are both gone, the proposition at issue for the 67.

Three Definitions of the Next Number, Written Down

The article has been using the phrase the next number in three different senses and they need separating, because the X-68A’s number is correct under one of them and not under the others. Let $\mathcal{A}_t$ be the set of design numbers allocated in a basic mission before a date $t$.

The first is one above the highest number ever allocated. The second is the lowest number never allocated. The third is one above the most recently allocated number, which differs from the first whenever an allocation has been made below the pointer.

The 2020 instruction’s procedure names the first, in the words last approved design number. The 1994 instruction named the second, in the words next available consecutive. The compiler’s published figures follow neither cleanly, because he excludes the allocations he judges out of sequence before applying the first. And his derivation for the FQ-48 follows the third, since the fighter numbers reach 117 while he reads the 48 as following the 47.

The second definition cannot be computed from any public document this article has found. That claim is narrower than saying that it cannot be computed from the public record at all. Evaluating the lowest never-allocated number over the register returns 1, because the register opens on 19 August 1998 and carries no row for the X-1 through the X-36. The founding document and the 2004 list reach further back and neither closes the gap, because each lists the designations current at its date of issue rather than every one ever allocated [AFR 66-11 and AR 700-26 and BUWEPS 13100.7, 18 September 1962] [DoD 4120.15-L, 2004 with Change 1]. A lowest-never-allocated number needs a complete history of allocations, and no public document is one.

Of the two definitions that are computable, both gave 67 on the day the X-68A was approved. At 2025-08-20 the highest research design number allocated was 66 and the most recently allocated was 66, carried by the X-66A, so one above either is 67.

Under either computable reading of the instruction the X-68A should have been the X-67A, and the three readings above exist to account for that.

His convention is also not the maximum, and that shows the figure to be a judgement. The register carries an X-76A, allocated on 20 October 2025, so the highest research design number is 76 and one more than the highest would be 77 [Parsch, DOD 4120.15-L Addendum]. He says 69 rather than 77, because he records the X-76 as out of sequence on the agency’s own account that the number is a deliberate nod to the revolutionary spirit of 1776. The published next-available number is therefore one more than the highest number he judges to have been allocated in sequence, and the judgement about which allocations are in sequence is his.

He applies the same convention consistently across three series, and consistency is the test of whether it is a convention or an error. In the unmanned series he states the next available number as 31 while the register carries unmanned numbers as high as 170. In the fighter series he states that the last sequentially assigned designation was F-23, that technically the next available number is therefore F-24, and that the series has to all intents and purposes stopped having a sequence [Missing USAF and DOD Aircraft Designations]. The convention is one rule applied with judgement, but not the rule the instruction states.

The two sources a requester could consult disagree about what to ask for, the instruction read literally giving 77 and the compiler 69, and only one of them is a published instruction while only the other is reachable.

The Official Source the Procedure Requires Does Not Resolve

The procedure’s research step depends on a source, and the instruction that states the procedure is the same instruction that cancelled the source. The instruction states this twice in its text and once in its summary of changes.

Major changes include the new site for the most recent MDS database on data.af.mil and the cancellation of DoD 4120.15-L, Model Designation of Military Aerospace Vehicles as the publically accessible MDS database.

DoD 4120.15-L, Model Designation of Military Aerospace Vehicles, will no longer be maintained as the primary source of approved MDS designators and popular names. However, other Federal Agencies and the public may obtain the latest version from data.af.mil.

The instruction also names where a Department component should look instead, a page on the Defense Technical Information Center’s DoDTechipedia site, and says that other Federal Agencies and the public may view or obtain an electronic tabular list by visiting the same address [DAFI 16-401].

The list’s Change 1 does not cancel anything, although the compiler records that the Department announced the list’s retirement in 2018. DoD 4120.15-L of 12 May 2004 incorporating Change 1 of 31 August 2018 was read from a public web archive snapshot, the Department’s portal having refused every client. Its summary of changes reassigns the office of primary responsibility to the Under Secretary of Defense for Acquisition and Sustainment and says nothing else, and the words retire, cancel and data.af.mil appear nowhere in it [DoD 4120.15-L, 2004 with Change 1].

So the list’s retirement is an act of the Air Force instruction and not a 2018 act of the Department list, and it came in two steps. The issue of 16 May 2014 already stated that DoD 4120.15-L would no longer be maintained as the primary source of approved designators, named the DoDTechipedia listing of the Defense Technical Information Center as the primary source, and sent the public to the National Technical Information Service for the last updated version [AFI 16-401, 16 May 2014]. The 2020 issue added the cancellation as the publicly accessible database and the address data.af.mil, and the article cites the instruments rather than a description of them.

That address does not exist. A query for the name against the authoritative nameservers for the af.mil zone returns no answer and a statement of authority, which is how a nameserver says that the name is not in the zone. A sibling name in the same zone resolves and so does a name in an adjacent zone, www.af.mil returning an address through a content delivery network and data.mil returning one directly, so the zone is being served and the query is reaching it. This was checked on 2 October 2026 from one host against the zone’s authoritative servers, so it is not a local resolver failure.

That result supports a narrower reading than it first appears to, and the article states the narrow version. A Department network commonly resolves internal names that the public Internet does not, so the measurement does not show that authorised users cannot reach the database. What it shows is that the official source the request procedure requires is not reachable at the address its cancellation notice named, from outside. That is a fact about public access, and it leaves the designation system’s internal function untouched.

The consequence for this article is a chain of four steps. The procedure requires research into the last approved design number. The official published list was cancelled. Its named successor does not resolve publicly. So the de facto public source is one private compiler’s reconstruction, obtained under the Freedom of Information Act, applying a next-number convention the instruction does not state. Every public claim about why the X-67 is empty passes through that reconstruction, and so do this article’s measurements.

The Department’s own portal still refuses the document, and the article records the refusal. The address for DoD 4120.15-L on the issuance portal returns HTTP 403, the Hypertext Transfer Protocol’s refusal status, to every client this article has tried [DoD 4120.15-L at the Department’s portal]. The document was read from a public web archive snapshot instead, and the October 1998 edition’s front matter from the compiler’s scan of it [DoD 4120.15-L, 2004 with Change 1] [DoD 4120.15-L, October 1998].

The 1998 edition shows what the public access used to be, the measure against which the present state should be read. Its foreword states that the list is approved for public release with distribution unlimited, and gives three routes to a copy, being a Department web address, the Defense Technical Information Center for registered users, and the National Technical Information Service for other Federal agencies and the public [DoD 4120.15-L, October 1998]. Three published routes in 1998, and in 2020 one address that does not resolve. The directive above the instruction still directs the Secretary of the Air Force to publish a single Department-wide source document of approved designators and names, and to publish it annually [DoD Directive 4120.15, 29 November 2004].

Sizing From First Principles

No X-67 vehicle was ever built, so there is no vehicle to size, and the opening already states this. The design number 67 was consumed by an allocation in the unmanned series, and nothing in the research series was ever drawn, specified or dimensioned under it [Parsch, DOD 4120.15-L Addendum]. This section records what the record supports about the vehicle that took the number and stops.

The record sizes the XQ-67A barely at all. The register’s only physical entry for it is an engine cell reading 1 Williams FJ44-4A, one of the official fields on a row whose description is not official wording [Parsch, DOD 4120.15-L Addendum]. No span, weight, thrust, endurance or speed appears in the register or in the releases.

What the record offers about the vehicles intended is comparative rather than dimensional, being the contrast between the two species quoted under The Aeroplane Was Built to Deny the Premise. That is a sizing question stated and not answered, and no document in the record answers it.

Two derivations in this article stand in for sizing, and neither sizes an aeroplane. The first, set out under A Designation Is a Code and the Instruction Gives Its Fields, sizes the designation field, which holds 23,976 distinct number-and-series codes per basic mission. The second sizes the economics of a shared chassis, in the part-count, unit-cost, penalty, break-even and refresh relations within What the Data Changed, and every parameter it uses is assumed because the shared fraction, the sharing penalty and the development costs are not public.

Dependent Systems

A designation that was never allocated depends on nothing, so the systems this section can describe are those of the aeroplane that took the number. The X-67 has no airframe, engine or avionics to list. The XQ-67A has a few, and the record names them without dimensioning any of them.

The propulsion entry is the only system the register records without the compiler’s mark. The XQ-67A’s row records 1 Williams FJ44-4A in an unmarked cell, while its stated purpose carries the mark saying the words are the compiler’s [Parsch, DOD 4120.15-L Addendum]. The airframe is described only as a common core chassis, which the laboratory calls a genus and to which mission kits such as an off-board sensing station or an off-board weapon station would be added [AFRL first flight release] [GA-ASI first flight release].

The mission systems appear only through the July 2025 flight test described under The Aeroplane That Took the Number, in the Record’s Own Words, and the releases name no sensor, computer or datalink by type [GA-ASI autonomy release].

The designation itself also has dependent systems, and they are documentary. The request procedure depends on an official source for the last approved design number, the instruction that states the procedure cancelled the public list, and the successor address it named does not resolve publicly [DAFI 16-401]. That chain is argued under the heading The Official Source the Procedure Requires Does Not Resolve.

The Flight Test Record

Nothing has flown as the X-67, and nothing can, because the number is unrequestable under the instruction in force. The register carries no research-series row with the design number 67, and the argument under Why the Slot Is Empty, So Far As the Record Says shows why the number cannot now be issued [Parsch, DOD 4120.15-L Addendum] [DAFI 16-401]. The flight test record of the X-67 is empty and will remain so.

The aeroplane that took the number has a short public record, being two reported flight tests in three releases, the first flight of 2024-02-28 and the autonomy flight of July 2025, both described under The Aeroplane That Took the Number, in the Record’s Own Words [AFRL first flight release] [GA-ASI first flight release] [GA-ASI autonomy release].

No flight data are public. The releases report that test points were completed and validated and give no measured value for any of them. The record does not say how many flights have been made in total, only that the first was the first of a series.

Comparison With Ground Prediction

For the X-67 there is no comparison to make, because no vehicle carried the number and none flew under it. No wind tunnel test, analysis or specification was ever produced for an X-67. The comparisons the record does support are of other kinds, and each is argued where it arises.

The aeroplane’s claim is public and its flight data are not. The laboratory states that the XQ-67A proves the genus approach and the contractor that it validated the genus and species concept, and no predicted or measured figure is published for any flight quantity, so the validation is reported and not shown [AFRL first flight release] [GA-ASI first flight release]. The commonality arithmetic under What the Data Changed is a model run on assumed parameters and not a prediction this aeroplane could have tested.

The designation procedure gives both an expectation and an outturn. Read literally, it would have produced an X-67A on 2025-08-20 and the record returned an X-68A [Parsch, DOD 4120.15-L Addendum] [DAFI 16-401], which the subsections headed Three Definitions of the Next Number, Written Down and Three Readings of the X-68A, and the Record Chooses None examine. The award record sets the contract values reported at award against their outturn, through the relative deviation $\epsilon$ displayed under What the Award Record Adds That the Releases Do Not, and that comparison shows that a competition was resolved rather than completed and shows nothing about how the aeroplane flew.

What the Data Changed, What the Anomaly Reveals About the Designation System

A skipped number is only interesting if the number was carrying information, so the question this section asks is what a design number indexes. The instruction gives an answer, the register contradicts it by measurement, and the aeroplane that consumed this number was built to prove that the instruction’s answer is the wrong shape.

A Designation Is a Code and the Instruction Gives Its Fields

The instruction defines the designator as two components separated by a dash. The first is alphabetic and describes the mission. The second is alphanumeric and describes the design number and the design series [DAFI 16-401]. A standard aircraft designator comprises an optional status prefix, an optional modified mission symbol, a required basic mission symbol, a required design number and a required design series. Only one symbol may be selected from each category. The register’s compiler sets out the same structure with its history in a separate article, which this one has read in part and does not follow where the instruction is available [Current Designations of U.S. Military Aircraft].

Write a designation as an ordered tuple of those fields.

Here $p$ is the status prefix, $q$ the modified mission symbol, $m$ the basic mission symbol, $d$ the design number and $s$ the design series letter. For the designation this article is about, the tuple is the experimental prefix, no modified mission symbol, the unmanned basic mission, the design number 67 and the series letter A.

The instruction bounds two of those fields numerically. Design numbers run consecutively from 1 to 999, and in the event that a design number reaches 999 the range expands to 9,999 and the character positions to the right of the dash expand with it [DAFI 16-401]. The series symbol follows a consecutive assignment starting with A, and the letters I and O are prohibited to avoid confusion with digits. So the series alphabet holds 24 letters rather than twenty-six, and the number-and-series field of one basic mission has the following capacity.

With $N_d$ equal to 999 and $N_s$ equal to 24 that is 23,976 distinct codes per basic mission, rising to 239,976 when the number field expands. The instruction also provides for the series letter to roll over, stating that at the end of the series symbol Z the next sequence advances the design number to the next consecutive unused number and begins again at A [DAFI 16-401]. That provision is the one place the instruction uses the phrase next consecutive unused number, and it concerns series exhaustion rather than requests.

The Instruction Says the Design Number Is an Ordinal Within One Mission

The definition is explicit, and everything below tests it. The design number is a required numeric symbol that identifies major design changes within the same basic mission, and the instruction illustrates it by stating that the 16 in F-16A represents the sixteenth approved designator for an aircraft with a fighter mission under the current reporting system [DAFI 16-401].

The series letter carries the next level down, and its criterion matters later. The series symbol identifies the production model of a particular design number and later models representing major modifications that significantly alter the aerospace vehicle’s systems, components, or change the logistics support of the vehicle [DAFI 16-401].

The system’s model of identity is a two-level hierarchy and the word major is undefined at both levels in the instruction now in force. A major design change earns a new number. A major modification earns a new letter. The instruction of 2020 does not say how to tell one from the other.

The Founding Document Did Say, and the Test It Gave Has Been Deleted

The instruction names three founding documents, and they are where its definitions began. The instruction states that the designator format was established by the Department on 18 September 1962 with the publishing of Air Force Regulation 66-11, Army Regulation 700-26 and Bureau of Naval Weapons Instruction 13100.7 [DAFI 16-401]. Those are one document issued three times over, a scan of it is public, and it has been read in full [AFR 66-11 and AR 700-26 and BUWEPS 13100.7, 18 September 1962].

It defines the design number in four words the later editions dropped.

Design Number. The sequence number of each new design of the same basic mission or type aircraft.

The founding definition does not say that a design number identifies major design changes. It says the design number is a sequence number. The increment test under The Instruction’s Claim Has a Testable Consequence and the Register Fails It was built to check that claim, and the two series it covers fail it in more than half of their steps, since only 24 of their 51 increments are exactly one, fifteen of 22 in the research series and nine of 29 in the unmanned series.

And where the modern instruction offers no test for what counts as a new design, the founding document offers three worked examples.

New design numbers will be assigned when an existing aircraft or airship is redesigned to an extent that it no longer reflects the original configuration or capability. Examples of changes requiring design redesignations on aircraft are as follows: (1) Changing the number of engines of a specific aircraft. (2) Changing the wing or control surface design of a specific aircraft from a straight wing to a swept or delta wing design. (3) Changing the empennage of a specific design from straight to swept surfaces or relocating the empennage.

All three are changes to the airframe’s shape or its propulsion. Not one of them is a change of mission equipment. The 1962 test is a test about the aeroplane and not about what the aeroplane is for.

That matters to this article’s keystone more than anything else in the founding document. A genus carrying several species shares the airframe and swaps the mission kit. Under the 1962 test that is not a new design at all, because the engine count, the wing and the empennage are the things held constant. The system that would have refused a new design number to a species is the system that wrote the definition the current instruction still paraphrases.

The series letter’s criterion, by contrast, has survived almost word for word. The founding document defines it as a letter denoting differences affecting methods of employment, differences affecting the relation of the vehicle to its ground environment, and major modifications resulting in significant changes to the logistic support [AFR 66-11 and AR 700-26 and BUWEPS 13100.7, 18 September 1962]. The logistic-support clause this article leans on is therefore sixty-three years old and not a recent form of words, and the 2004 list and the 2005 and 2020 instructions all carry it [DoD 4120.15-L, 2004 with Change 1] [AFI 16-401(I), 14 April 2005] [DAFI 16-401].

The two levels have aged in opposite directions. The lower level kept its criterion and the upper level lost its examples. The system can still say what a modification is and can no longer say what a design is.

And the Founding Document Defines Both of the Letters This Designation Carries

The X in XQ-67A is a status prefix and the X in X-67 would have been a basic mission, and the 1962 document defines both on facing pages.

X Experimental. Aircraft in a developmental, experimental stage where basic mission and design number have been designated but not established as a standard vehicle for service use.

X Research. Aircraft designed for testing configurations of a radical nature. These aircraft are not normally intended for use as tactical aircraft.

The first is the status prefix and the second is the basic mission and type symbol, and the borrowing this article is about takes a number from the series the second letter governs and attaches it to a designation carrying the first [AFR 66-11 and AR 700-26 and BUWEPS 13100.7, 18 September 1962].

The other letter has moved, and the move made the designation possible. In 1962 the Q is a modified mission symbol, defined as a drone, meaning an aircraft capable of being controlled from a point outside the aircraft [AFR 66-11 and AR 700-26 and BUWEPS 13100.7, 18 September 1962]. A modified mission symbol cannot carry a design-number series of its own, because it sits to the left of a basic mission letter rather than replacing it. In the register today the Q is a basic mission symbol with its own numbering, so there is an unmanned series for a number to be out of sequence in. The designation XQ-67A could not have been written in 1962, and not because the aeroplane did not exist.

The Agency Used to Choose the Number, and Now the Requester Does

The article has been circling this structural change, and the founding document states the original arrangement in two sentences.

The requesting service will indicate the desired mission or type symbols. The assignment agency will assign the applicable design number and/or the applicable series letter.

In 1962 the requester named the mission and the agency chose the number. The instruction in force today tells the requester to research the next-in-series designator from the last approved design number and to request it [DAFI 16-401]. The responsibility for choosing a design number has moved from the office that keeps the sequence to the office that wants the number.

That change is the precondition for what the next subsection measures. A number chosen by the keeper of the sequence is the next one by construction and carries no information. A number chosen by the requester can be chosen for what it says, and the unmanned series now carries nine numbers that are not the next one.

The Instruction’s Claim Has a Testable Consequence and the Register Fails It

The sixteenth-approved-designator sentence makes a claim with a consequence, and the consequence can be checked. If the design number is the ordinal of the allocation within its basic mission, then the $k$th design number allocated in that mission is $k$ itself.

That form cannot be tested against this register, and saying why is half the measurement. The register opens on 19 August 1998 and most basic missions were already numbering by then, so the origin of each sequence is outside the document and no ordinal can be read off it. Differencing the claim removes the origin.

So the instruction’s definition implies that each new design number exceeds the previous new one by exactly one, and that implication survives an unknown starting point. Only first allocations enter the walk, because a second row on a number already allocated is a series letter or a modified mission on an existing design rather than a new design number.

One refinement is needed. A row whose series letter is B or later is evidence that its design number existed before the row, whatever the register carries, because the series letter counts models of a number that was already approved. Without that filter the walk would count the X-40B as a new design number, and so misread the register’s edge as a government decision.

With the filter the research series gives 23 new design numbers over 22 steps, of which fifteen are exactly one. The unmanned series gives 30 over 29 steps with nine exactly one. The share of increments equal to one is

The Increment and the Pointer Measure Different Things, and the Article Needs Both

An increment of one and a pointer advancing by one are not the same statement, and conflating them double-counts. Write the running maximum as the pointer, the quantity the allocation procedure reads, and define its advance.

The two differ exactly when a number below the pointer is allocated, and there the identity is immediate.

An allocation below the pointer makes its increment negative and inflates the next one by the same amount, so counting skips off the increment counts each out-of-order allocation twice. The pointer does not have that defect, and the numbers actually skipped are the advances in excess of one.

For the research series that sum is eleven, across five advances greater than one, with sixteen advances of exactly one and one of zero, over 22 steps. The obedience rate is therefore the share of advances equal to one.

That is 72.7 percent for the research series and 50.4 percent across the whole register, being 67 advances of one in 133 steps, with 43 allocations below the pointer.

And the research series is the best-behaved sequence in the register. That sharpens the finding. Among the three basic missions with at least twenty steps the rates are 72.7 percent for the X series over 22 steps, 51.5 percent for the M series over 33 steps and 44.8 percent for the Q series over 29 steps.

The X-67 was lost from the one numbering sequence in this register that mostly does follow the rule.

Every Research-Series Skip Has a Cause This Series Has Written About

The five advances greater than one are not a statistical tail. They are a list, and the list is this series’ table of contents.

And the one advance of zero is the one number that came back. The X-49A was allocated on 2003-05-23, 464 days after the X-50A had passed it, filling a gap the pointer had already crossed [Parsch, DOD 4120.15-L Addendum]. So the register contains exactly one research-series number that was skipped and later recovered, and why that number could be recovered and the X-67 cannot is the subject of The Rule Changed, and the Register Holds the Before-Case in the Same Series.

Six Pairs of Numbers in the Register Carry the Same Official Sentence

If the design number were an ordinal over designs, two different numbers should not describe the same design. The register can be asked. Stripping the note references, which differ between two rows by construction, the register’s 532 well-formed rows carry 490 distinct descriptions of twenty characters or more. Sixteen of those descriptions are shared by more than one designation, and six of the shared ones span more than one design number, the other ten being series letters of a single number, the use series letters exist for.

Those six are all adjacent pairs, and there are no others.

In five of the six the contractor is the only field that differs, and three of the six were allocated on the same day. In the remaining one the contractor is the same firm and the register offers nothing to distinguish the two numbers. Two of the six are research-series pairs, a third of the set from one basic mission out of twenty, and this series has already written about both of them [Parsch, DOD 4120.15-L Addendum] [X-Planes: Invocon X-64].

The reading of that table is narrow but sufficient. The official statement of purpose does not distinguish these pairs, so whatever distinguishes two design numbers in the register’s practice is not the stated purpose. In five of six cases it is the designing organisation.

The Two Candidate Readings Are an Injectivity and a Well-Definedness, and Both Fail

Stating the two readings as properties of maps makes the failures precise and makes them countable. Let $\mathcal{N}$ be the set of mission-and-number pairs the register carries, and let the register supply two maps out of it.

The first reading, that a design number indexes a purpose, requires the purpose map to be injective. Two different numbers must not carry the same official statement.

It is not injective and the failures are counted. Over 490 distinct descriptions of twenty characters or more, six are shared by more than one design number.

The second reading, that a design number indexes an organisation, requires the maker map to be well defined, a weaker demand than injectivity and a different one. It asks only that every row on one number name the same firm.

It is not well defined either. Of 235 mission-and-number pairs, 27 carry rows naming more than one leading firm, or 11.5 percent.

The two readings fail in different ways, and the article states both without choosing between them. The purpose map is a function and is not injective, so a number carries no more purpose than its neighbour. The maker map is not even a function, so a number does not carry a maker at all.

And a Design Number Can Survive a Change of Manufacturer, So the Pairing Was a Choice

If the design number tracked the designer, a change of designer would force a new number, and it does not. Of the register’s design numbers, 103 carry more than one row. 36 of those carry more than one distinct contractor string and 27 carry more than one distinct leading firm name.

That last count supports two shares, and each is given with its denominator.

The first is as a share of the 103 numbers that carry more than one row, which is the population in which a change of firm is even possible. The second is as a share of all 235 numbers, the figure the well-definedness condition above reports.

The clearest case, and an extreme one, is a missile rather than an aeroplane. The design number 88 in the missile series carries ten rows across the series letters A, D, E, E, E, F, G, G, H and J, and the contractor field across them names seven distinct leading firms, among them Texas Instruments, Raytheon, Northrop Grumman and a naval weapons centre [Parsch, DOD 4120.15-L Addendum]. One design number, ten series letters, and a succession of primes spanning decades.

The system can hold one design number across a change of manufacturer. The X-46A and the X-47A could have been the X-46A and the X-46B, and the X-63A and X-64A the X-63A and X-63B. They were not, and the choice was available. The design number indexes neither the stated purpose nor the designing organisation reliably. It indexes a request, and the instruction leaves undefined what a request is for.

A Chosen Number Carries Information and a Sequential One Carries None

This is the quantitative reason a skipped number matters at all, and it can be stated exactly. Let $\nu_k$ be the number a sequential allocator would issue at the moment of the $k$th allocation, which by the procedure is one more than the last approved design number of that basic mission.

If every allocation takes the next number then $d_k = \nu_k$ for all $k$, the sequence is a deterministic function of its history, and a designation carries no information beyond its order. The quantity that carries anything is the departure.

An in-sequence allocation has $\delta_k = 0$. The information a chosen number can carry is bounded by the logarithm of the count of positions available to it, which runs from the sequential number to the top of the field.

The zero is better written as a conditional entropy than asserted, because in that form it describes the register, where an assertion would describe one allocation.

At the research series’ position in July 2023 that bound is 9.8657 bits, against 9.9643 bits for a number free to be anything. The bound covers almost the entire field because the field is almost empty. The earlier article made the same observation as an argument about scarcity, and here it serves as an argument about expressiveness. The system is not short of numbers, so a number is cheap, so choosing one can afford to say something.

The saying happens in the unmanned series, and the departures are measurable. Taking the sequential number as the top of the unbroken opening run, the convention the compiler’s published figures follow, the out-of-sequence unmanned allocations that carry a full date have the following departures.

The departures are 5 for the number 35, 17 for the number 48, 30 for the number 58, 37 for the number 67, 44 for the number 72 and 43 for the number 73, ranging from 5 to 44. There are six distinct values among six observations, so the empirical distribution is uniform and its entropy is 2.5850 bits, which equals the maximum 2.5850 attainable at that sample size.

That coincidence is a warning and not a result, and it is labelled as one. With six observations all distinct, the entropy estimate is pinned at its ceiling by the sample size and says nothing about the underlying distribution. The useful statement is the comparison of regimes rather than the number. A sequential series has zero bits of choice in every allocation. The unmanned series carries 69 rows and now has nine out-of-sequence design numbers against a sequential ceiling of 31, being 35, 42, 44, 48, 58, 67, 72, 73 and 170, of which six carry a full date and can be placed in time, being 35, 48, 58, 67, 72 and 73, and the register’s compiler has written of the fighter series that it has to all intents and purposes stopped having a sequence at all [Missing USAF and DOD Aircraft Designations]. The X-67 is one number lost in the middle of that transition.

The transition has a size in bits and summing the bound over the departures gives it.

That is 59.5355 bits across the six dated out-of-sequence unmanned allocations, or 9.9226 bits each. The figure is a ceiling and not a measurement of what was said, because a bound on what a channel could carry is not a reading of what went through it. What it establishes is the scale. A sequence that chose every number freely would carry about ten bits per allocation and a sequence that took the next one carries none, and the unmanned series has moved some distance from the second toward the first.

The Borrowing Claim Is Testable and It Passes

The compiler asserts that the 58 and the 67 came from the research series and gives no evidence for the second. The assertion has a testable consequence. If a borrowed number is taken from a neighbouring series’ queue, it should equal that series’ next number at the moment of the borrowing, and it should equal no other series’ next number, because a claim that matched many series at once would be worthless.

The test was run against every basic mission in the register at each out-of-sequence unmanned allocation’s date. There are twenty basic missions and six out-of-sequence unmanned numbers that carry a full date.

The test fires two times in six, on the numbers 58 and 67, and both times it names the research series and nothing else. It fires zero times on more than one series. It is silent on four numbers, 35, 48, 72 and 73, which are exactly the four the compiler either declines to derive or derives from somewhere other than a series queue.

A null model gives the coincidence a scale, and its reference class is stated here. Take the null to be that an out-of-sequence number is drawn uniformly from the two-digit integers above the unmanned series’ ceiling, the range 32 to 99 inclusive and so 68 values, written $B$. At the six dates in question the mean number of distinct next-in-series values falling inside that range is 3.3333, written $\bar{\tau}$, so the per-draw hit probability $\pi$ is 0.049020.

The expectation is the more legible quantity and it comes first.

At six draws and a per-draw probability of 0.049020 that is 0.2941 accidental hits expected, against two observed. At $n$ equal to six and $k$ equal to two the tail probability is 0.03159, or about one chance in 31.7. That is a weak signal on its own and the article does not pretend otherwise. It is not on its own. Both hits name the same series, the compiler independently states that series for both, and the research series independently shows a gap at both numbers. Four facts that could each be a coincidence are not four coincidences.

The test is silent on the 48 because the fighter numbers reach 117, so the compiler’s reading of it as following the F-47 uses the most recently allocated number, as noted under Three Definitions of the Next Number, Written Down [Missing USAF and DOD Aircraft Designations]. For the research series in July 2023 the highest allocated number, the top of the contiguous run and the most recently allocated all gave 67, so the X-67 case is clean and the 48 case is not.

The Aeroplane Was Built to Deny the Premise

Here the airframe earns its place in an article about a number. The design number exists to mark a major design change within a basic mission, so the system presumes that an aeroplane has a design identity and that changes to it are either major or minor. The programme that consumed the X-67 slot existed to show that an aeroplane’s identity is divisible.

The laboratory’s words, quoted above, put it as a taxonomy of a genus with mission kits added to the frame, and the contractor’s put it as a first instance [AFRL first flight release] [GA-ASI first flight release].

The laboratory also states what the two species were meant to differ in, and that detail sharpens the collision. The sensing station was conceived as slower while carrying sensors, with longer endurance. The weapon station was conceived as faster and more manoeuvrable, with less endurance and better range. The programme wanted to design both and to establish how much of the two could be made common [AFRL first flight release].

The designation question and the engineering question are the same question asked by two institutions that cannot hear each other. The instruction asks whether a change is major enough to earn a new design number. The programme asks how much of two aeroplanes can be the same aeroplane. Neither has a criterion the other would accept.

By the Instruction’s Own Test a Species Belongs on a Series Letter

The instruction’s only concrete criterion for the lower level is the one the genus concept is designed to defeat. A series letter marks later models representing major modifications that significantly alter the vehicle’s systems, components, or change the logistics support of the vehicle [DAFI 16-401].

A shared chassis is built precisely so that the logistics support does not change across species. The laboratory’s stated benefit is leveraging standard substructures and subsystems in the manner of an automotive product line, a claim about shared spares, shared tooling and shared training as much as about shared drawings [AFRL first flight release]. By the instruction’s own test, two species on one genus differ less than the instruction’s threshold for a series letter, let alone for a design number.

The register’s practice points the same way. The missile design number 88, described above, holds ten series letters across seven leading firms, so the system tolerates very large change under one number. Yet the first species on this genus received a design number of its own, taken from a different series.

And the system has shown in the same decade that it can extend itself, so the absence of any mechanism for a shared chassis is a choice rather than an oversight. The 2020 instruction’s summary of changes records one addition to the designator alphabet, a status prefix e meaning digitally developed, defined as aircraft that are engineered in a virtual environment [DAFI 16-401]. It is the only lowercase symbol in the scheme.

In the same issue that cancelled the public list, the system added a letter for how an aeroplane was engineered and added nothing for whether it shares a chassis with another aeroplane. The genus and species approach was already a laboratory programme with contracts when that instruction was signed [AFRL first flight release] [USAspending]. The system encoded the design environment and left design sharing unencoded.

The prediction is explicit and the record has not yet tested it. If an off-board weapon station is built on the same genus, the instruction’s criterion and the register’s practice both suggest it should be a series letter on the number the first species holds. The register’s post-2018 behaviour, in which a new programme name attracts a new number, suggests it will get its own. The weapon station does not appear in the register as last updated on 23 August 2026, so the question is open, and this article states which way it expects the answer to go and why it may be wrong.

What a Shared Chassis Does to a Part Count

The commonality claim is quantitative and its simplest form is a part count. Take a family of $S$ species, each of which would have $P$ parts if designed alone, and let $C$ of those parts be common across the family. The family then holds the common parts once and the distinct parts $S$ times.

Against wholly bespoke designs, which hold $SP$ distinct parts, the ratio is the quantity that matters, and writing the shared fraction as $\phi = C/P$ reduces it to two variables.

The limits are the test of the expression and both are right. At $\phi = 1$ every part is shared and $\rho = 1/S$, so a family of three species holds a third of the distinct parts, which is 0.3333. At $\phi = 0$ nothing is shared and $\rho = 1$. At a shared fraction of three fifths the ratio is 0.7000 for two species and 0.6000 for three.

The gradient is constant, so in this model the first shared part is worth as much as the last one.

At three species that is -0.6667, so each percentage point of sharing removes two thirds of a point of the distinct-part count. That constancy is the model’s weakness, and the penalty subsection below repairs it.

The relations in this section are the simplest defensible forms of a theory with a much larger literature. The modern treatment of modularity as a design strategy with an economic value is Baldwin and Clark’s, whose chapters on what modularity is and on the value of splitting and substitution set out the option-value argument this section does not attempt [Baldwin and Clark, Design Rules Volume 1] [What Is Modularity?] [The Value of Modularity, Splitting and Substitution]. Those three are cited from their registry records and not from a reading, as the reference definitions state, and the rest of that literature sits in the survey’s modularity cluster.

That relation is about drawings, qualification and spares rather than about unit cost, because a part that exists once has to be designed once, tested once, catalogued once and stocked once. It is also the weakest form of the claim, because it counts parts rather than weighting them, and the parts a family finds easiest to share are not the ones that cost most to design.

What It Does to a Unit Cost

Commonality acts on unit cost through quantity, and the standard relation is a power law. Let unit cost fall with cumulative quantity $Q$ as a negative power, and let the progress ratio $r$ be the factor by which unit cost falls when quantity doubles.

An eighty-five percent progress ratio gives $b$ equal to 0.23447 and a ninety percent ratio gives 0.15200. Now split each aeroplane into a shared fraction and a bespoke fraction. The step the argument turns on is that the two fractions are built at different quantities, and once it is written down the rest is arithmetic.

The bespoke fraction is built at the quantity of one species and gains nothing. The shared fraction is built at $S$ times that quantity and slides down the curve by $S^{-b}$. The two add in proportion.

At a shared fraction of three fifths, three species and an eighty-five percent curve, $\kappa$ is 0.86375, a saving of 13.63 percent. Two species give 0.91000 and a shared fraction of four fifths gives 0.81833. Writing the relation as a single slope makes every later argument an argument about one negative number.

For three species on an eighty-five percent curve the slope $a$ is -0.22709, and for two species it is -0.15000. The gradient of the cost factor is the slope itself, because this model is linear in the shared fraction.

At three species that is -0.22709. Because $a$ is negative for every $S$ greater than one, this model predicts that commonality always helps, monotonically. But it does not help without limit, and the limit matters because it bounds every claim the programme could make.

So no family, however large, saves more in unit cost than the fraction it shares.

At a three-fifths shared fraction the ceiling is 60.00 percent at any family size, against 13.63 percent reached at three species. The ceiling also makes the inversion well posed. Asking how many species a stated saving needs means solving the cost factor for $S$, and the solution exists only below the ceiling.

A tenth of unit cost needs 2.176 species at a three-fifths shared fraction, a fifth needs 5.637, and three tenths needs 19.227. The last figure is the useful one. A programme wanting a thirty percent unit-cost saving from commonality alone, at these parameters, needs about nineteen species on one chassis, and the programme that consumed the X-67 named two.

And the monotonic conclusion is still false, for a reason the model omits.

The Omission Is the Penalty, and Adding It Changes the Answer Qualitatively

A shared part is not a free part. A structure sized for the heaviest species is too heavy for the lightest. A chassis sized for the fastest is too draggy for the most enduring. The laboratory’s description of the two species it wanted, quoted above, names this trade [AFRL first flight release]. So write a penalty factor on the delivered cost of capability and ask what happens.

Take the penalty proportional to the shared fraction first, because it is the obvious choice and it gives a degenerate answer.

The curvature settles it in one line.

That is negative for every admissible pair, being -0.13625 at three species and a penalty coefficient of 30 percent, so the quadratic is concave and its stationary point is a maximum rather than a minimum. The cheapest shared fraction is therefore at one end of the interval or the other, and full commonality beats none exactly when the product at $\phi = 1$ falls below one.

For three species on an eighty-five percent curve that threshold is 29.38 percent, and for two species 17.65 percent. Below the threshold, share everything. Above it, share nothing. A model whose answer is always all or nothing is telling the reader about its linearity and not about aeroplanes.

A penalty growing as the square of the shared fraction is the better shape and it gives an interior answer. The first parts a family shares are the ones it costs least to share, so the marginal penalty should rise rather than stay flat.

That first-order condition is a quadratic in $\phi$ and its discriminant is the quantity that decides whether an interior optimum exists at all.

So an interior optimum exists only when the penalty coefficient clears a threshold set by the slope alone.

That is 15.47 percent for three species on an eighty-five percent curve and 6.75 percent for two. The threshold rises with the family size, and its limit follows from the slope’s.

So a sufficiently large family tolerates a sharing penalty of 300 percent before an interior optimum appears at all, the formal version of the observation that commonality is worth more the more species carry it. Below that the minimum is the corner at full commonality.

At a penalty coefficient of 30 percent the optimum is interior and the three cases separate properly. With $a = -0.22709$ at three species, the optimum and the two corners evaluate as follows.

The optimum shared fraction is 44.63 percent, at which the effective cost factor is 0.95235, a saving of 4.76 percent against sharing nothing. Full commonality at the same penalty gives 1.00479, which is 0.48 percent worse than sharing nothing at all. So at that penalty the interior optimum beats both corners and the corner the first model would have chosen is the wrong one.

Every number in the last three paragraphs uses assumed parameters and none of them is measured for this aeroplane. The shared fraction of the genus is not public, the sharing penalty is not public, and the number of species is one. What the relations establish is the shape of the argument and where it fails, the most a model fitted to a programme that has published no cost data can yield.

Inverting the Break-Even When One Species Exists

The programme’s claim is about development cost rather than unit cost, and a separate break-even applies. Let a bespoke aircraft type cost $c_b$ to develop, let the genus cost $\mathcal{C}_g$ once and let each species on it cost $c_s$.

That relation cannot be evaluated here, because $c_b$ is unknown and $S$ is one. Inverting it is the only direction the data supports.

Reading the genus cost as the platform-sharing programme’s total of 24,479,631.19 dollar and the first species at the Off-Board Sensing Station award of 67,986,112.00 dollar, the figures follow.

So a second species repays the chassis only if a bespoke demonstrator would have cost at least 80,225,927.59 dollar, which is 18.00 percent more than the first species cost. On the narrower reading that only the winning firm’s platform-sharing award of 10,674,331.00 dollar built the chassis, the requirement falls to 73,323,277.50 dollar at two species, a premium of 7.85 percent, and 71,544,222.33 dollar at three, a premium of 5.23 percent.

The sensitivity of the break-even to the chassis cost is a reciprocal and it gives the figure a programme manager would want.

At the two-species figures above that is one additional species required for every 12,239,815.59 dollar added to the chassis. The relation also states what the inversion shows, that the break-even is sensitive to a quantity nobody has published.

A third reading is that the chassis cost is inside the species cost and not separable at all, since the award record carries no line for a chassis. Under that reading the data cannot answer the question and the article states that rather than choosing the reading that gives a number.

Why Frequent Refresh Is Worth Something, Stated as a Shortfall

The programme’s second claim is about time rather than money and it can be made exact. The stated benefit is more opportunities for frequent technology refresh, against a complaint that the time to initial operating capability for military aircraft has been increasing [GA-ASI first flight release] [AFRL first flight release]. Let the available capability at the frontier grow exponentially and let a design hold the value it had when it was frozen.

A design frozen at the start of an interval of length $T$ delivers $F_0$ throughout, so the ratio of what it delivers to what is available is $e^{-g t}$, and the time average of that over the interval has a closed form.

The shortfall $\sigma$ has the right limits and a useful small-argument form. As $T$ tends to zero it tends to zero, as $T$ grows without bound it tends to one, and for small $gT$ the expansion is linear in the freeze interval.

Taking a two-year doubling of available capability gives $g$ equal to 0.34657 per year, from the relation below, the only place a rate enters.

The table shows the monotonicity at four values, and the derivative extends the claim to every interval.

That is positive for every positive argument, so the shortfall increases with the freeze without exception. It is 0.08940 per year at a three-year freeze and 0.01238 at fifteen, a ratio of 7.218, so the first year of delay costs about seven times what the fifteenth does. That is the quantitative form of the programme’s argument for cadence.

For a long freeze the exponential has died and the shortfall has a one-term form.

At fifteen years that approximation gives 80.76 percent against the exact 80.87 percent, so the asymptote is already within a tenth of a point at the longest interval the table carries.

Shortening the freeze from fifteen years to three takes the shortfall from 80.87 percent to 37.82 percent, and the gain is the difference.

That is 43.05 percentage points. That is the quantitative content of the refresh claim and it is much larger than the unit-cost saving commonality buys, which at the parameters above was 13.63 percent. If the genus concept is worth what the laboratory says it is worth, the value is in the cadence rather than in the parts.

The doubling time is assumed and nothing in the record supports it. A different doubling time scales every figure in the table, and the shortfall for a slowly improving capability is small at any freeze interval. The relation is exact and its input is a guess.

The Firm on Both Sides of the Gap

One register fact closes the section and it needs no arithmetic. The XQ-67A was allocated to General Atomics on 2023-07-19, and the X-68A, the allocation whose approval made the X-67 unrequestable, was allocated to General Atomics on 2025-08-20 [Parsch, DOD 4120.15-L Addendum]. The same firm is on both sides of the gap.

The register holds sixteen rows for that firm, being eleven in the unmanned series, four in the unmanned-control series and one in the research series. The one research row is the X-68A. For scale, the row counts by leading firm name over all 539 rows run 88 for Boeing, 84 for Lockheed, 78 for Raytheon, 33 for Northrop, 24 for Sikorsky and 16 for General Atomics.

No intent is claimed, since a firm holding this much of the unmanned register will appear on both sides of most things, and whether that allocation’s requester regarded the 67 as taken is a question the record does not answer.

Symbols

Every symbol the article uses, in one place. Where two meanings compete for one letter, the resolution is noted below the table.

The collisions, and what yielded. The sequential number takes $\nu_k$ so that the costs $c_s$ and $c_b$ keep the cost letter. The status prefix keeps $p$ and the null model’s probability takes $\pi$. The common part count keeps $C$, so the genus cost and the cluster set take calligraphic letters. The freeze interval keeps $T$ and the null model’s mean target count takes $\bar{\tau}$. The discriminant takes $\Theta$ because $\Delta$ is a difference throughout, and a row inside the well-definedness condition takes $w$ because $r$ is the progress ratio. $K$ counts the records the research survey admits while $k$ indexes an allocation, so the case is doing work. The distinct-part ratio $\rho$ and the capability ratio $\bar{\varrho}$ render similarly at small sizes and never share an expression.

Capability has no unit, because the refresh derivation needs only the ratio of two capabilities, and a designation’s fields have none, because they are symbols in a code, and only for that reason can a design number be chosen rather than measured.

The Contemporary Literature

The survey’s most useful finding is that this article’s literature is not an aeronautical literature. The X-67 is a number, the aeroplane that consumed it was built to prove a method of building aeroplanes, and the method has a large, quantitative and continuously active literature under product families, component commonality and modular architecture. That literature lives in mechanical engineering and management science, so most of the survey comes from a journal index rather than from a report archive.

The survey’s 1,109 research records fall across fifteen clusters, one cluster to each record. A record’s title is matched against the characteristic phrases of every cluster, and the record is filed in the first cluster in table order whose phrases its title contains, except two titles on product-line requirements engineering, which the requirements phrase alone would file under digital engineering and which belong with the software product lines under product families. A title may contain the phrases of more than one cluster, and the mean number of clusters a title matches says how often.

That is 1.0649 clusters per record across 1,181 matches, and 72 titles match two. A mean barely above one says the clusters are nearly disjoint, and that reflects the admission rule rather than the field, since each cluster admits on compound nouns the others do not use.

The shape of that table is the article’s subject restated. The three largest clusters are product families at 298, commonality at 207 and modularity at 161, which together are the engineering literature of when a thing is a new thing. The configuration identification and identifier clusters, at 59 and 63, are the literature of how engineered items and records are identified and named. The numerical taxonomy cluster at 24 is the literature of classifying at all.

The smallest cluster is attritability at four records, deliberately, because attritability is the X-58 article’s keystone and that article surveyed it, as noted under Out of Scope.

Two clusters hold literatures outside product design, and both are deliberate. The identifier cluster reaches persistent identifier systems, object identifier registries, namespace definitions and software versioning, because the questions a registry faces about allocating and retiring names are the questions the designation system faces and the answers are better documented. The numerical taxonomy cluster reaches the method of numerical classification, because the programme chose a biological metaphor and the discipline it borrowed from has a worked method for deciding what counts as a kind.

The oldest records in the survey show that standardisation is a far older subject than the designation system, and that naming entered the research record later. The survey reaches a paper on the standardisation of electrical engineering plant from 1900 and its discussion [Sellon 1900] [Crompton et al 1900], and a run of papers and section discussions on standardisation as applied to production engineering from 1933 [Standardisation as applied to 1933] [Wright 1933]. The question of how to standardise manufactured things is older than the designation system this article is about by two generations, and the engineering literature treated it as a subject in its own right before the first aeroplane flew. The question of how to name what engineers make reaches the survey’s research records in 1962, with a compilation of aerodynamic nomenclature and axes systems [Wright 1962], the year in which the designator format was established [DAFI 16-401].

Commonality

The sharing of parts and its measures. Commonality indices, component sharing across a product line, the inventory and assembly consequences of a shared part, and the standardisation of production, its older relative. 39 of these titles name standardisation and 30 concern inventory, stock or assemble-to-order systems, because a part shared across products pools the demand for it.

207 records. [A Product Family Design 2008] [Agrawal and Cohen 2001] [Aiers et al 1933] [Akkermans and van der Horst 2002] [Alizon et al 2006] [Alizon et al 2006] [Alizon et al 2007] [Alizon et al 2009] [Allen 1933] [Arakawa et al 2004] [Ayton et al 1933] [Bagchi and Gutierrez 1992] [Baker 1985] [Baker et al 1986] [Barber et al 1978] [Bedford 1933] [Bernstein et al 2011] [Bick et al 2014] [Bick et al 2015] [Blecker and Abdelkafi 2007] [Boas et al 2012] [Bollapragada and Akella 1993] [Boysen and Scholl 2009] [Cameron and Crawley 2012] [Chatras and Giard 2016] [Chen et al 2009] [Cheung 2002] [Chiu et al 2020] [Chowdhury et al 2012] [Chowdhury et al 2013] [Collier 1981] [Collier 1982] [Collier 1984] [Collins et al 1991] [Cooke et al 1933] [Cooke et al 1933] [Creighton et al 1987] [Crompton et al 1900] [Cronie and Thompson 1980] [Cunningham 1981] [Dang and Nagarur 2000] [Davis et al 1986] [Desai et al 2001] [Diamond 1951] [Dillon et al 1998] [Discussion on “The standardisation 1923] [Dong and Frank 2006] [Eilmus and Krause 2013] [Eynan and Fouque 2005] [Eynan and Rosenblatt 1996] [Eynan and Rosenblatt 2007] [Farrell and Simpson 2001] [Farrell and Simpson 2003] [Felice 1985] [Fellini et al 2006] [Fisher et al 1999] [Franke and Hobbs 1993] [Garmilla Manzano et al 2026] [Gerchak and Henig 1986] [Gerchak and Henig 1989] [Gerchak et al 1988] [Gilbert 1933] [Groocock et al 1933] [Guerrero 1985] [Hamdi and Tadj 2019] [Han et al 2019] [Heese and Swaminathan 2006] [Heikal et al 2019] [Henderson and Higer 2007] [Heradio-Gil et al 2011] [Hernández et al 2015] [Hidayat and Fadilla 2025] [Hidayat et al 2017] [Hill 1973] [Hillier 1999] [Hillier 1999] [Hillier 2000] [Hou and Lai 2015] [Ho¨ltta¨-Otto 2005] [Huang et al 2008] [Huang et al 2018] [Hölttä-Otto et al 2008] [Insulation commonality assessment phase 1973] [Izui et al 2009] [Jans et al 2008] [Jennings et al 2025] [Jiao and Tseng 2000] [Jiao et al 2007] [Johnson and Kirchain 2010] [Johnson and Kirchain 2013] [Jung and Simpson 2015] [Jung and Simpson 2016] [Jönsson and Silver 1991] [Kane] [Kannan and Nolte 1998] [Kariya et al 2007] [Khajavirad and Michalek 2007] [Khajavirad et al 2008] [Khire et al 2008] [Khire et al 2013] [Kim and Chhajed 2000] [Kim and Kim 2020] [Kim et al 2013] [Kim et al 2020] [Kim et al 2021] [Kima and Chhajed 2001] [Kline and Bachman 2006] [Kosanke and Nell 1999] [Kosanke and Vernadat 1998] [Kota and Sethuraman 1998] [Kota et al 1998] [Kourtis et al 2026] [Kruse and Gries 2011] [Kwapisz et al 2019] [Labro 2004] [le Maistre 1933] [Li Qian and Zhenyu Kong 2008] [Liu 2016] [Majerus et al 1999] [Marion et al 2007] [Mayall 1983] [Mirchandani and Mishra 2002] [Mohebbi and Choobineh 2005] [Moskovich 1978] [Nagarur and Azeem 1999] [Navarro Gonzalez 2022] [Nicol and Hooper 1999] [Ong et al 2006] [Pandey et al 2011] [Pasche and Sköld 2012] [Peng Chew et al 2006] [Pfeifer 2019] [Pillai 1956] [Pinson and Ryan 1972] [Pudge 1933] [Qian 2011] [Qian 2011] [Qiu et al 2018] [Ratcliffe 1965] [Rieker and Haraburda] [Rieker and Haraburda 1989] [Rojas Arciniegas and Kim 2010] [Rojas Arciniegas and Kim 2011] [Rojas Arciniegas and Kim 2012] [Rubio et al 2014] [Rutenberg 1971] [Sellon 1900] [Shamsuzzoha and Helo 2009] [Sheu and Wacker 1997] [Shi and Zhao 2014] [Shooter and Cohen 2010] [Siddique et al 1998] [Simpson 2017] [Simpson and D’Souza 2002] [Simpson and D’Souza 2004] [Sitcharangsie et al 2021] [Smith 1933] [Smith 2006] [Smith et al 2003] [Song and Zhao 2009] [Standardisation as applied to 1933] [Stojanovski and Selva 2017] [Study of the commonality 1975] [Su et al 2010] [Subramanian et al 2013] [Takai 2018] [Takai 2019] [Takai and Sengupta 2016] [Takai and Sengupta 2017] [Takeda et al 2018] [Tanabe and Awano 2019] [Taylor 2009] [Thevenot and Simpson 2004] [Thevenot and Simpson 2006] [Thevenot and Simpson 2006] [Thevenot and Simpson 2006] [Thevenot and Simpson 2007] [Thevenot and Simpson 2007] [Thevenot et al 2005] [Thevenot et al 2007] [Thomas 1992] [Thonemann and Brandeau 2000] [Tsubone et al 1994] [Van Dyke et al 1990] [Vance 1984] [Verhoef et al 2012] [Wan-Chuan Chan and Arıkan 2020] [Wang] [Wang et al 2011] [Wazed et al 2008] [Wazed et al 2010] [Wazed et al 2011] [Wee and Dada 2010] [Wettig 2002] [White 1978] [Wong et al 2019] [Wright 1933] [Xu et al 2020] [Yao et al 2017] [Ye and Gershenson 2009] [Yeager 1987] [Yong Lin 2006] [Yongjie et al 2021] [Zelm and Doumeingts 2013] [Zhang et al 2015] [Zhang et al 2019] [Zhou and Li 2007]

Product families

The design of families of products on a shared platform. Platform selection and optimisation, commonality and variety trade-offs, mass customisation and the management of product variety, with 31 titles on software product lines, the same idea in a discipline that builds variants from shared code.

298 records. [A Framework on Customer 2021] [Abdullah et al 2012] [Abu-Matar and Gomaa 2011] [Achatz 2011] [Advances in Product Family 2014] [Akundi et al 2005] [AlGeddawy 2014] [AlGeddawy and ElMaraghy 2011] [Alizon et al 2008] [Allada and Lan 2002] [Allison et al 2006] [Allon et al 2010] [Alsawalqah et al 2013] [America et al 2000] [Andersen] [Anderson 2013] [Aoki and Staeblein 2018] [Baylis et al 2018] [Berenbach] [Bibi] [Bin et al] [Birrenbach 2000] [Bonev et al 2015] [Bortolini et al 2023] [Bowman 2006] [Broy 2011] [Burge et al 2011] [Buschmann and Schwanninger 2011] [Böckle 2005] [Chandrasekaran et al 2004] [Chao and Ishii 2004] [Chen and Wang 2008] [Chen et al 2012] [Chen et al 2023] [Cheng and Zhu 2013] [Cheng et al 2015] [Choma Neto et al 2017] [Chong et al 2009] [Chowdhury and Siddique 2009] [Chowdhury et al 2011] [Chowdhury et al 2013] [Clements and Weiderman 1998] [Corbett and Rosen 2004] [D’Souza and Simpson 2002] [D’Souza and Simpson∗ 2003] [Cunha Reis] [Dai and Scott 2004] [Dai and Scott 2004] [Dai and Scott 2005] [Dai and Scott* 2005] [Dashofy and van der Hoek 2002] [de Lange 2002] [De Lit and Delchambre 2003] [De Lit and Delchambre 2003] [De Lit and Delchambre 2003] [De Lit and Delchambre 2003] [De Lit and Delchambre 2003] [de Weck 2006] [de Weck et al 2003] [Design of Product Platform 2008] [Dinnus and Pohl 2005] [Dolan and Lewis 2008] [Dolan et al 2000] [Du] [Du et al 2002] [Du et al 2012] [Economic models of product 2007] [Eggink] [Eggink and Groll 2020] [Eichstetter et al 2015] [ElMaraghy and AlGeddawy 2013] [Emmatty and Sarmah 2012] [Evaluating Product Family Development 2009] [Fan et al 2010] [Farrell and Simpson 2008] [Farrell and Simpson 2009] [Fellini et al 2006] [Finch 1999] [Fujita 2013] [Gao et al 2009] [Garneau et al 2013] [Gedell and Johannesson 2012] [Generic product model of 2025] [Gonzalez-Zugasti and Otto 2000] [Goswami and Allada 2007] [Greve et al 2019] [Hallsteinsen et al 2004] [Halman et al 2006] [Han et al 2019] [Harland and Uddin 2014] [He et al 2021] [He et al 2024] [Hendrickson et al 2008] [Holmqvist et al 2006] [Hu et al 2008] [Huang et al 2012] [Hume and Rosen 2012] [Jaring and Bosch 2004] [Ji et al 2013] [Jiang 2009] [Jiang and Allada 2001] [Jiang and Allada 2005] [Jiao] [Jiao 2012] [Jiao and Tseng 2000] [Jiao and Zhang 2006] [Jiao et al 2006] [Johannesson and Claesson 2005] [John and Villela 2008] [Johnson Lim et al 2011] [Jose and Tollenaere 2005] [Jung and Simpson 2018] [Jung et al 2017] [Kamrani 2013] [Karagoz et al 2021] [Khadke and Gershenson 2008] [Khajavirad and Michalek 2007] [Khajavirad and Michalek 2008] [Khajavirad et al 2013] [Kim] [Kim and Moon 2017] [Kim and Moon 2017] [Kim et al 2003] [Kim et al 2010] [Kim et al 2016] [Knauber and Thiel 2002] [Kokkolaras et al 2006] [Korhonen and Mikkonen 2004] [Krause and Eilmus 2011] [Krause and Gebhardt 2023] [Krause and Gebhardt 2023] [Krause et al 2013] [Krishnan 2008] [Krueger 2008] [Krueger 2011] [Krueger 2015] [Ku 2014] [Kuang and Jiang 2009] [Kumar et al 2004] [Kun and Dong 2006] [Kwak and Kim 2013] [Lamine et al 2005] [Lancaster 1998] [Lee et al 2025] [Lei] [Levandowski et al 2015] [Li 2010] [Li and Weiss 2011] [Li and Wu 2011] [Lim] [Lim et al 2010] [Lin 2004] [Liu 2012] [Liu and Hsiao 2005] [Liu et al 2009] [Luo et al 2013] [Lyons et al 2012] [Ma and Kim 2014] [Ma and Kim 2015] [Maccari] [Maccari 2002] [MacDuffie et al 1996] [Mannion and Savolainen 2014] [Martin and Ishii 2000] [Martin and Ishii 2002] [Martinez-Larrosa and Siddique 2026] [Martini and Forcellini 2015] [Matinlassi] [Messac et al 2000] [Messac et al 2013] [Model-Based Assessment of Variety-Induced 2025] [Model-Driven, Aspect-Oriented Product Line 2009] [Modrak and Bednar 2016] [Modrak et al 2014] [Mohagheghi and Conradi 2004] [Mohr 2002] [Moon and Simpson 2010] [Moon and Simpson 2013] [Moon et al 2009] [Moon et al 2011] [Moon et al 2013] [Morales 2013] [Morales 2013] [Multiple-View Requirements Models for 2009] [Nace and Koopman 2001] [Nakao and Iino 2021] [Nanda et al 2005] [Nayak et al 2000] [Nayak et al 2002] [Nelson et al 1999] [Nelson et al 2000] [Nidamarthi and Karandikar 2006] [Niemelä 2005] [Ninan and Siddique 2013] [Ninan et al 2024] [Nomaguchi et al 2004] [Nørgaard et al 2025] [Oh and Hong 2024] [Oh et al 2022] [Oh et al 2026] [Ohvanainen and Hietikko 2012] [Okpoti et al 2019] [Ostrosi Egon et al 2014] [Otto et al 2013] [Otto et al 2016] [Overview of GBOM Formation 2016] [Pakkanen et al 2016] [Pandit and Siddique 2004] [Park 2021] [Park and Shin 2007] [Park and Simpson 2006] [Park et al 2008] [Ping Jiang et al 2007] [Pirmoradi and Wang 2011] [Pirmoradi et al 2013] [Pohl and Reuys 2002] [Product Platform and Product 2006] [Rai et al] [Raudberget et al 2014] [Riaz et al 2017] [Robert] [Roger Jiao et al 2007] [Rosen and Choi 2021] [Savolainen and Mannion 2012] [Schmid and Lee 2013] [Schuh et al 2017] [Schulze and Li 2009] [Shah et al 2011] [Shao et al 2012] [Shatnawi et al 2017] [Shooter 2006] [Siddique 2006] [Siddique and Adupala 2005] [Siddique and Adupala 2008] [Siddique and Rosen 1999] [Simpson 2004] [Simpson 2006] [Simpson et al 1999] [Simpson et al 2001] [Simpson et al 2001] [Simpson et al 2006] [Simpson et al 2010] [Simpson et al 2011] [Siu] [Software and systems engineering] [Software Product Line Engineering 2009] [Software Product Line Engineering 2012] [Sopadang et al 2002] [Sotos et al 2013] [Staples and Hill] [Steuer and Whitcomb 2003] [Stewart and Yan] [Sundgren 1999] [Taguchi Integrated Real-Time 2010] [Tang et al 2017] [Thevenot and Simpson 2009] [Thiel and Hein 2002] [Trask and Roman 2008] [Tseng and Jiao 1998] [Tucker 2013] [Uddin et al 2018] [Universal product family design 2012] [Voelter and Visser 2011] [Vogt et al 2023] [Wallmark] [Wang] [Wang et al 2011] [Wang et al 2012] [Wang et al 2018] [Wang et al 2025] [Wei and Qin] [Weingärtner 2002] [Wenlei Zhang and Yushun Fan 2006] [Wijnstra 2002] [Wijnstra 2004] [Windheim 2019] [Windheim 2019] [Windheim 2020] [Wu and Li 2011] [Xiao 2012] [Xu and Jiao 2013] [Yang et al 2004] [Yang et al 2005] [Yao et al 2023] [Ye and Gershenson 2008] [Yue Yang and Xibin Wang 2009] [Zacharias and Yassine 2007] [Zha and Lu 2002] [Zhang 2012] [Zhang 2014] [Zhang and Cheng 2005] [Zhang and Tseng 2007] [Zhang et al 2007] [Zhao 2010] [Zhao et al 2022] [Zheng et al 2021] [Zhu 2004] [Zhu 2006]

Modularity

The architecture of a product divided into modules. Module identification, modular product architecture, the theory of modularity and its option value, and the dominant-design and architectural-innovation work that explains when an architecture becomes fixed. 65 of these titles name the design structure matrix, which is the field’s principal instrument for deciding where a module boundary falls.

161 records. [Abdelsalam et al 2014] [Adrees 2021] [Akoto Chama and Bertram 2018] [AlGeddawy and ElMaraghy 2015] [AlGeddawy et al 2017] [Alizon et al 2006] [Amalia et al 2018] [Ananno et al 2026] [Anderson and Tushman 1990] [Asaga and Nishigaki 2022] [Asan and Polat 2008] [Aurich et al] [Baldwin and Clark 2000] [Baldwin and Clark, The Value of Modularity, Splitting and Substitution, chapter in Design Rules Volume 1, The Power of Modularity, MIT Press, 2000, registry record only] [Baldwin and Clark, What Is Modularity?, chapter in Design Rules Volume 1, The Power of Modularity, MIT Press, 2000, registry record only] [Bao et al 2025] [Benkoczi et al 2018] [Borjesson and Hölttä-Otto 2012] [Borjesson and Sellgren 2013] [Boudouh and Grunder 2015] [Brovar and Fortin 2024] [Brown 2004] [Browning 2013] [Börger and Schulte 2000] [Carrascosa et al 1998] [Chantzaras 2024] [Chen Chun-Hsien et al 2013] [Cheng et al 2019] [Chitsaz and Efatmaneshnik 2023] [Choudhuri et al 2019] [Clark and Baldwin 2002] [Cook and Coates 2016] [Correa et al 2017] [Dahmus et al 2000] [Dahmus et al 2001] [Dambietz 2021] [Domercant 2017] [Dunbing Tang et al 2009] [Eppinger and Browning 2012] [Ertas and Gulbulak 2021] [Essers] [Ezzat et al 2019] [Fan et al 2024] [Farooq et al 2013] [Fischetti 1974] [Fujita 2002] [Fujita et al 1999] [Wahdan et al 2016] [Gao and Zhang 2020] [Gasnakis and Homberg 2013] [Gauthier] [Gershenson et al 2003] [Gershenson et al 2004] [Gong et al 2017] [Gravina da Rocha et al 2019] [Greve and Krause 2018] [Gualberto] [Gunawan 2012] [Gupta 2021] [Hamidreza Kasaei et al 2010] [Hata et al] [Heberlig and Palaoro 1971] [Helmer et al 2010] [Helo 2006] [Henderson and Clark 1990] [Hohnen et al 2013] [Hossain et al 2021] [Houshmand and Valilai 2012] [Huang and Wang 2025] [Huang et al 2010] [Huang et al 2012] [Huang et al 2025] [Hölttä-Otto et al 2012] [Introduction to Design Structure 2012] [Jelenić 2024] [Jiang et al 2026] [Jung and Simpson 2016] [Kashkoush and ElMaraghy 2017] [Kim and Moon 2016] [Kim et al 2019] [Kim et al 2025] [Kim et al 2025] [Kimita et al 2021] [Kässi et al 2008] [Lane 1990] [Lange and Imsdahl 2013] [Larsson and Åslund 2001] [Lee 2012] [Li 2023] [Li and Chen 2008] [Maldonado et al 2017] [Marion et al 2014] [Matos et al 2007] [Meyer et al 2014] [Milev et al 2009] [Ming Dong et al 2010] [Modrak and Soltysova 2021] [Monetti et al 2025] [Monetti et al 2026] [Morris and Steiner 2006] [Murmann and Frenken 2005] [Murmann and Frenken 2006] [Nelson 2017] [Newcomb et al 2003] [Niutanen and Otto 2017] [Pektaş and Pultar 2006] [Pinto 2019] [Piran et al 2020] [Pisch et al 2024] [Plant 2017] [Product Development by Using 2015] [Qiao et al 2017] [Qiao et al 2018] [Qiu 2010] [Reynerson] [Reynerson 1999] [Ribeiro et al 2007] [Roh et al 2025] [Rosen 1996] [Samarasinghe et al 2019] [Sanaei et al 2015] [Shamsuzzoha 2011] [Shamsuzzoha and Helo 2010] [Shamsuzzoha and Helo 2012] [Shamsuzzoha et al 2020] [Shamsuzzoha et al 2023] [Shekar et al 2011] [Shikata et al 2013] [Shoval et al 2016] [Smit and Pistorius 1998] [Stettner 2000] [Sullivan et al 2001] [Sun and Lau 2019] [Sun et al 2016] [Tang 2008] [Tang 2008] [Tang et al 2008] [Tang et al 2009] [Tang et al 2017] [Tang et al 2017] [Teacher et al 2023] [The Modular Operators 2000] [Tokunaga and Fujimura 2013] [Ulrich 1994] [Use Modular Architecture] [Uusitalo 2014] [Wallance 2021] [Wallance 2021] [Wan 2014] [Wang and Tseng 2011] [Wang et al 2023] [Windheim 2019] [Xiaogang et al 2006] [Yan and Feng 2011] [Yang et al 2014] [Yang et al 2016] [Yee Soon Lim 2007] [Yen and Smith 2009] [Zamani and Beigzadeh 2023] [Zeng et al 2025] [Zou et al 2026]

Open architecture

Open interfaces as an acquisition policy. 32 of these titles name the modular open systems approach, and the rest concern open architectures in defence and integrated modular avionics, which is the government’s version of the commonality question asked about interfaces rather than parts.

47 records. [ade et al 2017] [Agarwal 2026] [Blank et al 2025] [Bonilla-Ortiz and Verma 2020] [Boudreau 2007] [DeMinico et al 2022] [Department Of Defense Washington Dc 1998] [DiVito 1998] [DuBois 2022] [DuBois et al 2021] [Farr et al 2024] [Ford and Dillard 2008] [Ford and Dillard 2009] [Gaska 2007] [Gaska 2012] [Holmstedt 2023] [Jacobs 2024] [Jakovljevic 2006] [Karimi and Werner 2014] [Kovach et al 2021] [Lyke et al 2019] [DeMinico et al] [Matthews and Dubois 2021] [Melber et al 2015] [Motzet et al 2006] [Optimization Techniques for Multi-Core 2025] [Padilla et al 2025] [Peters et al 2022] [Pierce and Littlefield-Lawwill 2008] [Plonsky and Herber 2026] [Reinhart et al 2008] [Rendon 2006] [Rendon 2007] [Rendon 2008] [Roberts et al 2023] [Rose et al 2014] [Schrage 2023] [Shaver et al 2016] [Siddapureddy 2011] [Space And Naval Warfare Center Arlington Va 1996] [Srivastava and Rice 2014] [Toppler et al 2025] [Vogler et al 2026] [Watza and Cooper 2023] [White 2023] [Zimmerman et al 2018] [Zischka and Finnegan 2023]

Cost of variety

What it costs to make an aircraft, and to make several kinds. Cost-estimating relationships for aircraft, cost growth in defence acquisition, economies of scope and the cost of variety, and 32 titles on the learning curve on which the unit-cost derivation above rests.

56 records. [Aircraft Cost Considerations 2010] [Asher and Maggelet 1980] [Asher and Shafer 1985] [Attia et al 2016] [Badiru 2021] [Beltramo and Anderson 1977] [Beltramo et al 1979] [Camm et al 1987] [Chatzimichali and Tourassis 2008] [Cody and Tiedje 1996] [Cody and Tiedje 1997] [Cost-Estimating Relationships for Aircraft 1972] [Cui and Weston 2010] [Cunningham 1980] [Curtis 1987] [Donath et al 1981] [Ettlie and Penner-Hahn 1994] [Gansler et al 2010] [Gray 1979] [Gruber 1992] [Gruber 1994] [Guangming Chen and McLennan] [Gudmundsson 2014] [Gudmundsson 2022] [Harofteh and Saljooghi 2023] [Hogan et al 2020] [Hogan et al 2022] [Horowitz et al 2014] [Jones 2018] [Kim and Lee 2018] [Kneece et al 2014] [Learning Curves Production Breaks/Lost 2015] [Learning Curves Production Breaks/Lost 2026] [Lu et al 2008] [Lundberg 1956] [Majd and Pindyck 1987] [Mazzola and McCardle 1997] [McClure 1977] [McNicol and Wu 2014] [Myrodia et al 2021] [Nemet 2006] [Pananiswami and Bishop 1991] [Parrish 1978] [Pedersen and Slepniov 2016] [Pindyck and Majd 1996] [Ridder 2024] [Smunt 1999] [Smunt 2011] [Sokri and Ghanmi 2017] [Stahl et al 1984] [Teng et al 2014] [Towill 1985] [Using the Learning Curve 2010] [Why do Aircraft Cost] [Why do aircraft cost 1999] [Wong 2013]

Flexibility

The value of being able to change a design later. 17 of these titles concern real options in engineering systems design, and the rest concern technology insertion and the management of obsolescent parts, which is the problem frequent refresh addresses.

28 records. [Barriers to Real Options 2009] [Berkowitz and Rowlands 2005] [Caputo and Cardin 2021] [Caputo and Cardin 2021] [Cardin et al 2021] [Cardin et al 2022] [Cardin et al 2023] [Guinivan 2004] [Identifying Real Options to 2009] [Koo 2013] [Krystallis et al 2026] [Ma et al 2017] [Maddux 1999] [Mandelbaum et al 2008] [Mukhtar and Agarwal 2009] [Origins of Real Options 2009] [Peterson 1971] [Rahmat et al 2017] [Real Options Design for 2009] [Real Options in Engineering 2009] [Real Options in Engineering 2009] [Real Options in Manufacturing 2009] [Regan 2001] [Rennenkampf and Rhyne 2004] [Underwood 2011] [Van den Broeke et al 2017] [Wearne 1992] [Xie and Cardin 2017]

Crewed and uncrewed teaming

Uncrewed aircraft fighting beside crewed ones. 75 of these titles name manned-unmanned teaming and 4 name collaborative combat aircraft, so this literature is a literature of the teaming concept, its workload and its interfaces.

81 records. [Albawaneh et al 2025] [Andrews et al 2020] [Andrews et al 2020] [Arthur 2008] [Athanasiadou et al 2026] [Bailey 2008] [Bajpai et al 2025] [Banas et al 2020] [Catak et al 2025] [Choi et al 2026] [Cosenzo 2010] [Das et al 2017] [Ding et al 2026] [Donkels and Voigt 2019] [Driggs et al 2026] [Dudek and Schulte 2022] [Flaherty et al 2006] [Flores et al 2025] [Friedrich and Schulte 2025] [Fu Li et al 2008] [Gozzi and Schulte 2026] [Hammarbäck 2022] [Hammarbäck et al 2023] [Han et al 2025] [Heilemann and Hollatz 2020] [Hoke 2018] [Hong and Baek 2025] [Hwang and Han 2025] [İşci and Günel 2021] [Jeong 2026] [Jeong et al 2026] [Jung et al 2025] [Kiam et al 2022] [Kim 2026] [Kim and Kim 2026] [Kim and Lee 2022] [Kohn 2022] [Kong et al 2024] [Korecki and Hoika 2026] [Kwon et al 2025] [Künzel et al 2025] [Lee and Jeong 2025] [Lee and Kang 2025] [Lee and Lee 2026] [Lee and Na 2023] [Lee and Park 2024] [Lee et al 2025] [Lee et al 2025] [Lewkowicz 2026] [Lyons et al 2021] [Maier et al 2024] [Malhotra et al 2024] [Meyer and Schulte 2020] [Meyer and Schulte 2022] [Michelson and Domercant 2019] [Minglang et al 2018] [Park et al 2022] [Paul et al 2025] [Pericon 2024] [Pletsch 2021] [Qi and Xie 2026] [Roncolini and Quaranta 2024] [Rosenblum et al 2022] [Roth and Schulte 2020] [Roth and Schulte 2022] [Roth and Schulte 2022] [Roth et al 2020] [Ruf and Stütz 2016] [Ruf and Stütz 2018] [Schmitt and Schulte 2015] [Schmitt and Schulte 2015] [Schwerd and Schulte 2021] [Shively et al 2011] [Sparrow and Henschke 2023] [Strenzke et al 2011] [The People’s Liberation Army’s 2025] [Voigt et al 2019] [Wuwer et al 2025] [Yang et al 2024] [Yoo et al 2021] [Yuan et al 2025]

Attritability

Aircraft cheap enough to lose. Attritable design trades, a low-cost rapidly built unmanned aircraft, damage tolerance of additively manufactured steels for attritable aircraft and a semi-expendable vehicle. The cluster is small by design, for the reason given above.

4 records. [Bateman et al 2007] [Colombi et al 2017] [Jones et al 2021] [Ma et al 2021]

Autonomy

The software that flies an uncrewed aircraft, and the trust placed in it. Autonomy architectures and their formal verification, human-autonomy and human-agent teaming, tactical datalinks and the interoperability of unmanned systems.

14 records. [Bloch et al 2023] [Bloch et al 2023] [Chen and Barnes 2013] [Hartel et al 1977] [Ho et al 2017] [Humphrey and Muñoz] [Prinzel et al] [Müller et al 2023] [Neubauer et al 2022] [Neubauer et al 2023] [Sperling et al 2008] [Varela 2014] [Wright et al 2022] [Wright et al 2024]

Configuration identification and nomenclature

How engineered items and their records are identified and named. 53 of these titles name configuration identification, management or control, and the rest concern engineering change and the nomenclature standards of engineering disciplines. No title in the survey names a designation system, so the nearest engineering literature to the register is the practice of identifying configurations, as the section headed Thin Subjects records.

59 records. [Appendix 1 Standard nomenclature 2019] [Barkstrom 2003] [Bright et al 1993] [Buckle 1982] [Calvo-Manzano et al 2001] [Carnduff 2004] [Change Control and Configuration 2008] [Change Control and Configuration 2023] [Change management, configuration management 2018] [Colston 1998] [Configuration Identification] [Configuration Identification 1995] [Configuration Identification 2004] [Configuration Identification 2004] [Configuration Identification for Digital] [Configuration Management and Software 2004] [Configuration Management Identification 2007] [Configuration-Management] [Dart 1992] [Department Of Defense Washington Dc 1993] [Dunleavy 1988] [Fabry 1980] [Feiler and Downey 1990] [Fonder and Headley 2013] [Harvey 1986] [Hoek et al 1995] [Hoek et al 1998] [Hoek et al 1998] [Hoek et al 1998] [Hood 1987] [How Configuration Identification Works 2004] [Joeris 1997] [Jucevic 1978] [Kögel 2008] [Kölmel and Eisenbiegler 2001] [MacKay 1996] [Martinis 1990] [Masmoudi et al 2017] [Mathur and Jain 2011] [Mehmood et al 2013] [Missouri Univ-Rolla 1988] [Piligian et al 1968] [Programme Management. Configuration Management] [Quantick 2001] [Searle 1977] [Sligh et al 2002] [Smith and Wandelt 1991] [System Configuration Identification 2005] [Tang et al 2017] [Tate 1998] [Tichy 2001] [Tomayko 1986] [Tomayko 1990] [Transportation Systems Center Cambridge Ma 1989] [Vines 1989] [Wagner and Foster 1990] [Wallnau 1992] [Wright 1962] [Young 1989]

Identifiers

How registries allocate, define and keep names. 23 of these titles concern persistent identifiers and 16 concern versioning schemes, together with the digital object identifier, object identifier registries, uniform resource name namespaces and the Legal Entity Identifier.

63 records. [Aghassibake et al 2023] [Ajibode et al 2025] [Bangert and Frances 2017] [Bellini et al 2014] [Bellur and Vallieswaran 2006] [Berber and Yahyapour 2017] [Carvalho 2023] [Ceravolo and Bellini 2010] [Chan and Milne 2013] [Chandrakar 2006] [Chen et al 2019] [Cox 2008] [Dappert et al 2017] [Decan and Mens 2021] [Digital Object Identifier DOI® 2009] [Digital Object Identifier DOI® 2011] [Digital Object Identifier system] [DIN 31646 2013-01, Information und] [Golovachev 2026] [Hamilton 2002] [Hardisty et al 2021] [Herb 2021] [Hewlett 2016] [Hewlett 2017] [Hoffman 2010] [Housley 2013] [Housley 2014] [Housley 2014] [Housley 2021] [Iman 2025] [IMF Participation in the 2013] [Information and documentation. Digital] [Information technology – Procedures] [Islam et al 2022] [Kennickell 2016] [Kim and Park 2004] [Lam et al 2020] [Lehnert et al 2019] [Lemieux 2014] [Lemieux 2016] [Li et al 2023] [Luo et al 2015] [Milne and Chan 2019] [Nasrullah et al 2025] [Ochoa et al 2022] [Opdebeeck et al 2021] [Paglialonga and Schirnick 2020] [Panter et al 2025] [Paskin 1999] [PID Service an 2015] [Pinckney et al 2023] [Raemaekers et al 2017] [Rzepa 2018] [Schmitt et al 2015] [Sharan 2019] [Thompson and Tong 2018] [van Veenendaal 2023] [Wang et al 2008] [Wang et al 2017] [Weigel et al 2014] [Yun 2019] [Zhang et al 2022] [Ziedorn 2020]

Numerical taxonomy

The numerical method for deciding what belongs with what. The principles and practice of numerical taxonomy, its methods for choosing characters and comparing classifications, and reviews of its founding books.

24 records. [D’Agostino and Russell 2014] [Barthélemy and Brucker 2000] [Corliss et al 1974] [D’Agostino and Russell 2005] [Fiala 1983] [Gould 1974] [Gyllenberg and Koski 1996] [Hubac 1983] [James 1964] [Leuschner 1981] [Maze 1966] [McNeill 1984] [Numerical Taxonomic Methods 1982] [Numerical Taxonomic Methods 2024] [Numerical Taxonomy 1983] [Numerical Taxonomy 2017] [Numerical Taxonomy and Cluster 1994] [Numerical Taxonomy. The Principles 1975] [O’brien and Colwell 1988] [Principles of Numerical Taxonomy 2013] [Quesne 1969] [Schlee et al 1975] [Singh et al 2002] [Sneath 1985]

Unmanned aircraft design

Designing and sizing an uncrewed aircraft. Conceptual design studies of unmanned combat and solar aircraft, design programmes and modular unmanned airframes.

19 records. [Alexis and Tzes 2012] [Altman 2008] [Cummings et al 2018] [Fry 2008] [Fujinaga et al 2007] [Gambaryan 2021] [Herbst and Klöckner 2014] [Karabetsky 2018] [Khreish et al 2005] [Kim and Oh 2017] [Kotarski et al 2021] [Neufeld and Chung 2005] [Sadraey 2010] [Sathe and Pant 2010] [Szafranski et al 2014] [Unmanned Aerial Vehicle Design 2024] [Wu 2023] [Yang et al 2013] [Zheng et al 2022]

Demonstrators

Aircraft built to show that something can be done. 13 of these titles name an X-plane, being the X-22A, X-31, X-33, X-37 and X-45A or the series as a whole, and the rest concern rapid acquisition and prototyping, technology readiness levels and the Orbital Space Plane’s demonstrators.

24 records. [Arellano et al 2015] [Austin and Rising 2000] [Austin et al 2000] [Boyce et al 2013] [Brinker 2004] [Bulkley and Davis 2013] [Chyma 2010] [Coleman et al 2015] [Cosentino 2008] [Ford et al 2012] [Haspert and Beauregard 1998] [Jacobson 2004] [Jacobson 2004] [Marchese 1963] [Powers and Schellenger 1989] [Rasch and Robert A. 2011] [Reichenfeld and Jones 1999] [Sauser et al 2009] [Schweikhard et al 1998] [Schweikhard et al 2001] [Smith 2004] [Sterk and McAtee 2015] [Turner 2003] [Wise 2003]

Digital engineering

Engineering a system through its models. Model-based systems engineering, virtual prototyping, digital twins and requirements engineering.

24 records. [Alexandru 2020] [Assef and Geiger 2023] [Black et al 1990] [Black et al 1990] [Blackburn 2014] [Blackburn et al 2014] [Bonanne 2011] [Darpel et al 2019] [Dougal and Monti 2011] [Eckert 2006] [Crues] [Ewing 2022] [Garcia et al 1994] [Hanratty et al 1999] [Holt and McNeillis 2007] [Hwang et al 2022] [Quinlan] [Mead 2003] [Orhan et al 2026] [Ortiz 1992] [Ozkaya and Turunc 2025] [Patria 2013] [Wallace et al 2001] [Young et al 2004]

Where the Framing Breaks Down

The title says the slot was taken by the XQ-67A, and the record supports a narrower statement than that. Taken implies an act upon the X-67, and the article’s finding is that nobody had to decide to skip it. The number 67 went to an unmanned allocation on 2023-07-19, and the X-67 became unrequestable only when the X-68A was approved on 2025-08-20 [Parsch, DOD 4120.15-L Addendum] [DAFI 16-401]. The vehicle consumed a number, and a later request closed a slot.

The derivation from the research series is an inference and not a documented fact. The compiler asserts it with the phrase just like the X-58 and gives no reasoning [Missing USAF and DOD Aircraft Designations]. The test in the passage headed The Borrowing Claim Is Testable and It Passes supports it, with a tail probability of 0.03159 that the article calls a weak signal on its own. No document states that anybody chose 67 for that reason.

The analogy with the X-58 breaks at the officiality line. The XQ-58A’s description is official Department wording and the XQ-67A’s is not, so the phrase just like the X-58 joins a case the government described to one it did not. That boundary is drawn under the heading The Second Case Falls on the Far Side of a Line the First Did Not.

The framing of the skip as visible at the article’s date also breaks, and the dated note under Epistemic State records how. The X-68A row reached the public register only after this article’s date, so a reader of the register at that date would have seen 67 as the next number rather than as a passed one.

And the aeroplane crosses the boundary the opening draws. The opening treats the XQ-67A as evidence about the number and not as the subject, yet the part-count, unit-cost, sharing-penalty, break-even and refresh derivations are about the aeroplane’s method. Every input to them is assumed, and the article states that their yield is the shape of an argument rather than a measurement.

The Source Base

This article rests on a narrow base and the section exists to say how narrow. Everything about the designation comes from one compiler’s reconstruction and one published instruction. Everything about the aeroplane comes from two press releases and the federal award record. No document in either group was written to answer the question this article asks.

The research survey draws on three archives, being the National Aeronautics and Space Administration’s Technical Reports Server, the Defense Technical Information Center and the Crossref bibliographic index, called below the reports server, the defence registry and the bibliographic index. It holds 1,109 research records, being 1,105 from those archives and four hand-written theory sources, of which 130 are report primaries, being 11.7 percent, by the identifier rule set out under The Definition of Primary. The median publication year is 2011. Of the 1,062 dated records, 371 were published in 2015 or later, being 34.9 percent, and 181 before 2000, and the range runs from 1900 to 2026.

The reference base as a whole is 1,198 definitions and it partitions four ways.

22 are hand-written references to the register, the instructions, the releases, the award record and the trade press, two of which the list marks as secondary, four are hand-written theory sources, 67 are the prior articles of this series, and 1,105 are research records from the three archives. Of the 1,109 research records the 979 that are not reports carry bibliographic-index identifiers, which is the proportion the subject dictates rather than a choice.

The Reports Server Holds Almost None of This

The reports server’s own commonality literature is mostly spacecraft, as the commonality row of the table in the next subsection shows, and six of the ten reports-server records in the commonality cluster concern space vehicles, stations or spaceflight hardware. How much two vehicles can share is the question the XQ-67A was built to answer whatever the vehicles are, so that government literature belongs to the subject, but the journal literature on product families and modular architecture is far larger and the reports server holds almost none of it.

That is a finding about the subject. Commonality and product family design are a manufacturing and management literature, and a research agency that builds aeroplanes publishes about aeroplanes. Whether the laboratory’s own programme documents exist in the report literature under other names is open, the Low Cost Attritable Aircraft Technologies initiative and the Low-Cost Attritable Strike Demonstrator being two such names, together with the Broad Agency Announcement the trade coverage dates to September 2020.

The Programme’s Own Words Are Unsearchable

Every word the laboratory and the contractor use for the programme collides with another literature, and searching the archives for each bare term shows it.

The bibliographic index was no better and in two cases was worse. There, platform returns platform capitalism, platform urbanism, the platform economy, a media journal and an oligonucleotide synthesis platform, with nothing engineering in ten. commonality returns a dictionary entry for the word itself, commonality in tinnitus perception and commonality of beliefs in economics. genus species classification returns ten biological results in ten, and common chassis aircraft family returns a fire-fighting truck, a passenger-boarding lift, automotive chassis grease and a glider undercarriage.

The acronym collision is the one worth recording for its own sake. The programme’s own abbreviation was already taken, inside the same agency, by the Space Shuttle’s Orbiter Boom Sensor System, and all ten of the reports server’s first results for it are Shuttle return-to-flight material. A programme named after a function will collide with every other programme named after a similar function, and this series has now met the same class in a carbohydrate and a bird.

The literature’s own vocabulary finds the subject where the programme’s does not. product family, component commonality, design structure matrix, modular open systems approach and product platform design return their own subject at close to ten in ten. The X-66 article reports the same lesson in the opposite direction, since there the report literature’s vocabulary was the aeronautical one and here it is not aeronautical at all.

An Address That Returns a 404 Page Under a 200 Status

The contractor’s page for its aircraft family is not a document. It answers with a success status and 20,974 bytes, and its body reads 404 Error, Page Not Found. A success status is not a document, and this page is a case in which the status and the content disagree.

The consequence for the article is substantive rather than cosmetic. Secondary coverage speculates that the XQ-67A represents the first member of a named family of aircraft the contractor has disclosed separately. Contemporary reporting names the contractor’s submission as the Gambit and reports the contractor saying that the selection would validate the Gambit series’ design concept [AFRL Chooses General Atomics Drone for Off-Board Sensing Station Program]. With the maker’s own page unreachable, the article asserts no tie to the family beyond that report, which is what the record supports.

What the Cited Addresses Return

Every hand-written address in the article has been fetched, and there are 26 of them. Twenty-one return a document, four are held by a publisher that declines automated clients and are instead confirmed against the bibliographic index by author and title, and one returns neither. The one that returns neither is the Department’s own cancelled list at its issuance portal, which answers HTTP 403 to automated clients, and the article says in its own definition that it was not read [DoD 4120.15-L]. A fetch that fails tells a reader nothing about whether a citation is correct, so each of the four is resolved against the index and the resolved title and authors are compared against the label the article prints, and all four agree.

The Definition of Primary

Primacy is a property of the document, and this survey derives it from the identifier. A research record is a report primary when its address is on the reports server or the defence registry, or when it carries a digital object identifier under the 10.21236 prefix, which Crossref registers for the defence registry’s reports. The report-primary share is the fraction of research records that are report primaries.

Of the 1,109 research records, 130 are reports held by the reports server or the defence registry, so $\varpi$ is 11.7 percent, split 37 from the reports server and 93 from the defence registry. The rule does not reach the documents this article actually rests on, being the register, the instructions and the releases, which are hand-written references rather than research records, so the share describes the survey and not the article’s evidence about the X-67.

Thin Subjects

There is no literature on the X-67 and the article states that rather than implying coverage. The survey holds no record whose subject is this designation, because a designation that was never allocated produces nothing. The 1,109 research records map the field the question sits in and not the question, and the only sources that speak to the X-67 directly are the register, the compilation of missing designations and the instruction.

Six subjects that this article’s derivations depend on have thin literatures. Each count below is the number of research records in this survey whose title carries at least one of the listed terms, with a hyphen read as a space. Where a row joins two groups of terms with the word together, a title must carry a term from each group.

The designation, the aeroplane and its metaphor have no literature at all. No research title names the X-67, the XQ-67A or the Off-Board Sensing Station, none names a designation system, and none carries genus, species or chassis. This is the same negative result the X-66 article reported for airport design and the X-65 article for its effector, and it has the same cause. A laboratory demonstrator whose purpose is a production method does not generate a research literature under its own name.

Commonality across a family of aircraft is a small part of a large literature. Of the 17 titles, 4 name an unmanned aerial vehicle and the other 13 concern crewed aircraft, most of them airliners, transports and commuter aircraft, so the shared-chassis argument above draws on a literature written about airliners and about products in general.

Technology refresh appears in one title, on technology insertion in evolutionary acquisition, so the refresh derivation stands on its own relations rather than on a published model of refresh cadence.

The allocation of identifiers appears in three titles, one of them the procedure by which a registry allocates algorithm numbers for the security extensions of the domain name system, which is the closest the survey comes to the designation system’s own allocation question.

What the Survey Admits

A record is admitted only when a person reading its title finds it on this article’s subject, and every admitted title also carries the characteristic phrases of at least one of the fifteen clusters, so no residual cluster is needed.

Many words this subject depends on also name something else, and a record that shares only such a word is excluded. Numerical taxonomy applied to plants, microorganisms, soils, wines and psychiatric diagnoses is another science using the method, so its applications are excluded while its method and theory are kept. Learning curve also names surgical, financial and trainee learning curves, standardisation also names the standardisation of cheese, Ayurvedic preparations, herbal products, surveying practice and medical devices, modularity also names the graph sense used in network science, commonality also appears in the thermal and fluid sciences and names a statistical technique for finding the shared tool behind defective semiconductor wafers, modular also names porous chemical frameworks and cognitive architectures, namespace also names an isolation mechanism of an operating system kernel and the declarations file of a software package, and demonstration also names a flight display team and an experiment flown on a parabolic flight. Nomenclature of crystal chemistry and of chemical substances names natural rather than engineered objects. Real options valued without an engineering object, including in organisational design, are a finance technique and are excluded, and so are product variety and economies of scale in trade theory. Records about a demonstrator or an unmanned aircraft that report only an aerodynamic, thermal, aeroelastic, handling, flight-control or calibration result are excluded, because the article uses demonstrators as programmes and not as test articles. Obituaries, conference session headers, meeting programmes, figure records and supplementary files are excluded as well, because a survey counts research works and those are not works or are parts of works. Reviews of books, dictionary and other reference entries of two pages or fewer, news and other notices, and a second deposit of a document already in the survey are not admitted either, while an authored encyclopedia article of more than two pages and a textbook chapter are research works and stay.

The theory of modularity is admitted by its own distinctive phrases, because its foundational works state the theory rather than an application of it and carry no engineering noun. Product modularity is admitted as a phrase in its own right for the same reason, since a core reference on its measures and design methods pairs the word with no engineering noun either [Gershenson et al 2004]. The cognitive, neural and graph senses of the word are excluded.

Some records of doubtful relevance are kept. Methodological papers on numerical taxonomy, identifier registries outside engineering such as the Legal Entity Identifier and a statewide student identifier system, learning curves in energy production, the nomenclature standards of engineering disciplines, and modular architecture applied to buildings, recipes and services each share a method or an information-infrastructure sense with this subject, so each is kept and a reader may weigh it accordingly.

Every research title has been read for relevance, so the off-topic share that remains is a matter of reading judgement. A reading of titles cannot show that a work is relevant in substance, so this is a statement about titles only.

Epistemic State

This article’s claims divide unusually sharply, because its primary sources are a table and an instruction and almost everything else is computed from them. One division below, Measured From the Documents Rather Than Asserted by Them, holds claims that are measurements of a document rather than assertions by it.

The Article’s Own Vantage Point

This article carries the editorial date 2025-12-12 and three of its statements are dated after that. All three are measurements the article made of itself rather than events in the subject. The register is read from its page as last updated on 23 August 2026, the hostname check against the authoritative nameservers was run on 2 October 2026, twice, and the reference survey’s year range reaches 2026 because the bibliographic index carries records published after the dateline.

A survey is written from current knowledge, and this section says where that knowledge postdates the article’s own date. No event in the subject is dated after the dateline. The register’s latest allocation used here is 2025-11-21, the aeroplane’s latest flight test is July 2025, and the latest instruction is of 3 November 2020.

The one place this matters to a reader is the hostname. A name that did not resolve on 2 October 2026 may have resolved on 2025-12-12 and may resolve later. The article’s claim is about the date it was checked and not about the whole interval, and no other claim here could change without any document changing.

A note on the register. The X-68A and X-76A rows first appeared in the public register between 15 January and 1 February 2026, after this article’s date. The register’s page as archived on 10 December 2025 [Parsch, DOD 4120.15-L Addendum, as archived on 10 December 2025] and on 15 January 2026 [Parsch, DOD 4120.15-L Addendum, as archived on 15 January 2026] ends its research rows at the X-66A, and the page as archived on 1 February 2026 [Parsch, DOD 4120.15-L Addendum, as archived on 1 February 2026] carries both new rows. In the register, the passed number at 67 is visible only through the X-68A row, so a reader of the register at this article’s date would have seen the X-67 as the next number rather than as a passed one. The skip is therefore a fact that the register made public after this article’s date, and the statement above that no event in the subject is dated after the dateline holds for the allocations and not for their publication.

Historical Fact, From Primary Documents

• The designation XQ-67A was allocated on 2023-07-19 to General Atomics, with USAF as sponsor and an engine entry of 1 Williams FJ44-4A [Parsch, DOD 4120.15-L Addendum].
• The X-66A was allocated on 2023-05-03 and the X-68A on 2025-08-20, the latter to General Atomics [Parsch, DOD 4120.15-L Addendum].
• The XQ-58A was allocated on 2017-07-10 to Composite Engineering [Parsch, DOD 4120.15-L Addendum].
• The XQ-67A first flew on 2024-02-28 at the General Atomics Gray Butte Flight Operations Facility near Palmdale, California [AFRL first flight release] [GA-ASI first flight release].
• The contractor was selected in 2021 to design, build and fly the aircraft and the laboratory initially worked with five industry vendors [GA-ASI first flight release] [AFRL first flight release].
• The federal award record carries two Off-Board Sensing Station contracts, of 67,986,112.00 dollar and 16,003,302.00 dollar, beginning 2021-10-26 and 2021-10-25, and five platform-sharing contracts totalling 24,479,631.19 dollar awarded in spring 2019 [USAspending].
• In July 2025 the aeroplane flew with government reference autonomy and an active tactical datalink [GA-ASI autonomy release].
• The governing instruction defines the next-in-series designator from the last approved design number of the same basic mission, refuses requests in reverse or skipped sequences, and reserves to the allocating office the authority to skip a design number at discretion [DAFI 16-401].
• The founding document of 18 September 1962 defines the design number as the sequence number of each new design of the same basic mission, gives three worked examples of changes requiring a new design number, and states that the requesting service names the mission while the assignment agency assigns the number [AFR 66-11 and AR 700-26 and BUWEPS 13100.7, 18 September 1962].
• That document also defines the X as both a status prefix meaning experimental and a basic mission symbol meaning research, and defines the Q as a modified mission symbol meaning drone rather than as a basic mission [AFR 66-11 and AR 700-26 and BUWEPS 13100.7, 18 September 1962].
• It publishes the next research design number as 22A, in a list of four current research aircraft [AFR 66-11 and AR 700-26 and BUWEPS 13100.7, 18 September 1962].
• The editions of 1994, 1997 and 2005 all direct the coordinating office to assign the next available consecutive design number, and none contains the word skip [AFJI 16-401, 9 September 1994] [AFJI 16-401, 1 September 1997] [AFI 16-401(I), 14 April 2005].
• The edition of 16 May 2014, which supersedes the 2005 one, defines the next designator from the last approved design number, refuses requests in skipped sequences for vehicles with an extensive history of designations, and states that skipped design numbers cannot be requested or used, and it states that DoD 4120.15-L will no longer be maintained as the primary source of approved designators [AFI 16-401, 16 May 2014].
• The Department directive of 29 November 2004, as amended by Change 2 of 17 September 2020, requires a unique designator for each vehicle and annual publication of a single Department-wide source document, and states no rule on choosing or skipping a design number [DoD Directive 4120.15, 29 November 2004].
• The 2020 instruction cancels DoD 4120.15-L as the publicly accessible database, names data.af.mil as where the public may obtain the latest version, and adds a status prefix e for aircraft engineered in a virtual environment [DAFI 16-401].
• Change 1 of 31 August 2018 to DoD 4120.15-L reassigns the office of primary responsibility and does nothing else, the words retire, cancel and the successor address appearing nowhere in it [DoD 4120.15-L, 2004 with Change 1].
• The October 1998 edition is approved for public release with distribution unlimited and gives three routes to a copy [DoD 4120.15-L, October 1998].
• The 2004 edition lists the X-41A, the X-42A and the X-43A as approved designators [DoD 4120.15-L, 2004 with Change 1].
• The instruction defines the design number as identifying major design changes within the same basic mission, and the series letter as identifying major modifications that significantly alter systems or components or change the logistics support of the vehicle [DAFI 16-401].

Measured From the Documents Rather Than Asserted by Them

This division exists because the article’s central claims are counts over a table. Each of these was computed by parsing the register and each can be recomputed by anyone who parses it the same way.

• The register carries 539 designation rows, 532 of them well formed, and of the well-formed rows 503 carry a full date and 29 a partial one, spanning 1997-12-10 to 2025-11-21. Of the 539 rows, 103 carry the compiler’s mark of some kind.
• There are 31 research-series rows carrying 25 distinct design numbers from 37 to 76, and fifteen of the 40 positions in that span carry no row.
• The count of research-series rows carrying the design number 67 is zero.
• 21 research rows are fully official, seven have an unofficial description, one is partly unofficial and two are entirely unofficial rows. The XQ-58A’s description is official and the XQ-67A’s is not.
• Six pairs of design numbers in the register carry identical official descriptions, and in five of them the contractor is the only differing field.
• 27 design numbers carry rows naming more than one leading firm, and the missile design number 88 carries ten series letters across seven firms.
• The research series has two companions in the unmanned series carrying the experimental prefix and nothing else, being XQ-58A and XQ-67A.
• The unmanned series carries nine out-of-sequence design numbers against a sequential ceiling of 31, with numbers reaching 170.
• The hostname the cancellation notice named as successor to the printed list returns no answer from the authoritative nameservers for its zone, checked on 2 October 2026, while a sibling name in the same zone and a name in an adjacent zone both resolve.

Engineering Analysis, Derived Here and Checkable

• The number-and-series field of one basic mission holds 23,976 distinct codes, rising to 239,976 when the design-number range expands as the instruction provides.
• A sequential allocation carries no information beyond its order, and a chosen one carries at most 9.8657 bits at the research series’ position in July 2023.
• The departures of the dated out-of-sequence unmanned allocations are 5 for the number 35, 17 for the number 48, 30 for the number 58, 37 for the number 67, 44 for the number 72 and 43 for the number 73. Their entropy is 2.5850 bits, which equals the maximum 2.5850 attainable at six observations, so the entropy figure is pinned by sample size and is not a result.
• The distinct-part ratio of a family is one over the species count plus the complement of that times the unshared fraction, giving 0.6000 at three species and a shared fraction of three fifths and 0.3333 in the limit of full sharing.
• With an eighty-five percent progress ratio the learning exponent is 0.23447, and the fleet-average cost factor at three species and a shared fraction of three fifths is 0.86375, a saving of 13.63 percent.
• With a sharing penalty linear in the shared fraction the optimum is always a corner, and full commonality beats none when the penalty coefficient is below 29.38 percent at three species.
• With a sharing penalty growing as the square of the shared fraction an interior optimum exists only when the penalty coefficient exceeds three times the square of the sharing slope, which is 15.47 percent at three species. At a coefficient of 30 percent the optimum shared fraction is 44.63 percent, saving 4.76 percent, while full commonality is 0.48 percent worse than sharing nothing.
• The time-average capability shortfall of a design frozen for an interval is one minus the ratio of one less the exponential of minus the growth-rate-interval product to that product, giving 37.82 percent at three years and 80.87 percent at fifteen on a two-year doubling.
• A chassis costing the whole platform-sharing programme repays itself over two species only if a bespoke demonstrator would have cost 80,225,927.59 dollar, a premium of 18.00 percent over what the first species cost.

Inference, Stated as Such

• That the number 67 was taken from the research series is an inference, now supported by evidence the register’s compiler does not cite. The number equalled the research series’ next number at the moment of allocation, it equalled no other series’ next number, and the same test fires for the only other case the compiler derives this way. The inference is strong but not a documented fact.
• That the X-68A’s approval made the X-67 unrequestable follows from the instruction’s refusal of requests in skipped sequences, applied to the register’s dates. The instruction is a fact, the dates are facts, and the conjunction is an inference.
• That the design number indexes a request rather than a design or a designer is an inference from the identical-description pairs and the multi-firm numbers. Both measurements are facts and the conclusion drawn from them is not.
• That a species on a shared genus falls below the instruction’s threshold for a series letter is an inference from the logistics-support clause and the article’s most contestable claim.
• That the designation arriving mid-build is the ordinary case for a modern demonstrator is an inference from one case and should be treated as a hypothesis the remaining articles can test.
• That the laboratory’s account of deciding at the end of 2021 is most likely an error for the end of 2022 or a reference to the award itself is an inference from a twelve-month base period, and the record contains nothing that chooses.
• That a species on a shared chassis would not have earned a new design number under the 1962 test is an inference from three worked examples, all of which concern the airframe’s shape or its propulsion and none of which concerns mission equipment. The examples are facts and the application is not.
• That the 1997 wording permitted the X-49A’s fill is an inference from a procedural duty on the assigning office rather than from a stated permission to a requester.

What the Record Does Not Settle

• Who caused the X-67 to be skipped, or whether anybody did. The three readings in the body differ in agency and agree in every observable. The record contains no allocation letter, no refusal, no exercise of the discretion clause and no statement of intent.
• Why the number 67 rather than any other. The compiler’s derivation is an analogy to the earlier case, the test in this article corroborates it, and no document says that anyone chose it for that reason.
• Whether the design-number pool is shared across basic missions in practice. The instruction says it is not. The register’s behaviour is consistent with its being so. No publication states the practice.
• Whether the successor designation database exists and is maintained. The address does not resolve publicly and a Department network may resolve it. This article cannot see inside.
• What the earlier Department directives behind the instructions say. DoD Directive 4505.6 of 6 July 1962, which the founding regulation implements, and DoD Directive 4120.15 of 2 May 1985, which the 1994, 1997 and 2005 instructions implement and the lists are reissued under, are named in documents this article has read and have not themselves been read. Their successor of 29 November 2004 has been read as amended in 2020, and it states no numbering rule [DoD Directive 4120.15, 29 November 2004].
• What the Off-Board Sensing Station cost to develop, or what any part of the genus cost. The award record gives contract amounts and the releases give none, and no document separates the chassis from the species.
• Whether a second species was ever built. The off-board weapon station is described as a concept the programme wanted to design and no such aircraft appears in the register as last updated on 23 August 2026.
• Whether the aeroplane is part of the family its maker has disclosed separately. Contemporary reporting names the maker’s submission as the Gambit and reports the maker saying that the selection would validate the Gambit series’ design concept [AFRL Chooses General Atomics Drone for Off-Board Sensing Station Program], and the maker’s page for that family returns a not-found document under a success status.
• How many species a genus would need to carry before the approach paid. The relation is derived here and every input to it is unpublished.

Out of Scope

Each exclusion below is a thing a reader could reasonably look for here and not find.

• The airframe’s performance. No span, weight, thrust, endurance or speed appears above, because the register records none of them and the releases record none of them. The aeroplane enters this article as evidence about a design number and the article declines to assemble a specification from trade coverage.
• Attritability and the arithmetic of losing aircraft. The X-58 article derived the cost per sortie, the break-even survival probability and the dependence on a cost ratio, and showed where the argument fails. That material is not repeated and the cluster in the survey is deliberately small.
• The collaborative combat aircraft programme. The register carries the YFQ-42A, the YFQ-44A and the YFQ-48A, all out of sequence, and the compiler’s derivations for them appear above only as entries in a table. Those aircraft are a different subject and two of them are not in the research series at all.
• The X-68A. It appears here as the allocation whose approval closed the slot and as a date. It is the next article in this series, and its programme, its contractor’s other work and its anomalies are treated there.
• Autonomy and crewed-uncrewed teaming as technical subjects. The 2025 flight test is cited for what it shows about the aeroplane’s role as a testbed. The autonomy stack, the datalink and the teaming concept are a literature this article surveys and does not analyse.
• The wider history of the designation system. The 1962 unification, the pre-1962 service systems and the popular-name rules are reached only where the instruction’s current text requires. The system’s history is left to the closing article of this series.
• The platform and commonality literature as an engineering discipline, including commonality indices with competing definitions, design structure matrix clustering and option-valuation methods. The survey points at that literature and the article does not summarise it.
• Whether the genus and species approach is a good idea. The article derives what it would have to be worth and reports that none of the inputs is public. It takes no position.

Conclusion

The X-67 does not exist, the number 67 does, and the register records the second with an analogy to the case before it. The compiler’s entry says that just like the X-58 the slot was skipped after the allocation of the XQ-67A, and that sentence is all the public reasoning there is [Missing USAF and DOD Aircraft Designations].

The analogy is sound and the article had to supply the evidence for it. The number 67 was the research series’ next number on the day the unmanned allocation took it, the number 58 was the research series’ next number on the day the earlier one took it, and across six out-of-sequence unmanned numbers tested against twenty basic missions the test fires two times and names the research series both times [Parsch, DOD 4120.15-L Addendum]. The compiler asserts the derivation and declines to say why. The register he compiled contains the reason.

What the earlier article could not say is who skipped it, and the instruction turns out to say that the question is malformed. The next designator to request is defined from the last approved design number, and requests in reverse or skipped sequences are not accepted [DAFI 16-401]. Nobody had to decide to skip the X-67. Somebody had to ask for the X-68, which happened on 2025-08-20, 763 days after the borrowing, and from that approval the X-67 became unrequestable. A skipped number is not a vacancy in this system. It is a position behind a pointer that only moves forward. The rule entered the instruction with the issue of 16 May 2014 [AFI 16-401, 16 May 2014], so the X-58’s slot had already closed the same way, and the X-67 is the second research number made unrequestable by it.

The second case differs from the first in the one respect a register exists to record. In 2017 the government stated what the aeroplane taking the number was for, and the register carries that statement as official wording. In 2023 it did not, and the register’s description of the XQ-67A carries the mark saying the words are the compiler’s. The date, the contractor and the engine remain official. So the government will say officially which engine is in the aeroplane that consumed the X-67 and will not say what the aeroplane is for, and the compiler’s phrase just like the X-58 reaches across that boundary.

And the aeroplane that consumed it was built to deny the premise the number rests on. A design number marks a major design change within a basic mission, which presumes that an aeroplane has one design identity. The Off-Board Sensing Station existed to show that it has two levels, a shared genus and a replaceable species, and the laboratory said so in those words [AFRL first flight release]. By the instruction’s own criterion a species belongs on a series letter, because the only concrete test the instruction offers for that level is whether the change alters the logistics support of the vehicle, and a common chassis is built so that it does not.

The founding document is harder on the point than the instruction now in force. On 18 September 1962 the Department defined the design number as the sequence number of each new design and gave three worked examples of what forces one, being a change in the number of engines, a change of the wing from straight to swept or delta, and a change or relocation of the empennage [AFR 66-11 and AR 700-26 and BUWEPS 13100.7, 18 September 1962]. All three are the airframe and none is the mission equipment. A genus holds those three constant and swaps what the 1962 test ignores. The system that wrote the definition would have refused a species a number of its own, and the system that inherited the definition without the examples gave one.

The register shows the same thing from the other side and shows it twice. Six pairs of design numbers carry identical official descriptions, and in five of them the contractor is the only field that differs, so the number does not index the stated purpose. Yet 27 design numbers survive a change of leading firm, one of them across seven firms and ten series letters, so the number does not index the designer either. What it indexes is a request, and the instruction never defines what a request is for.

The arithmetic of the method the aeroplane proved shows where its value would have to lie, and that place is not the parts. By the cost factor $\kappa$ under What It Does to a Unit Cost, at three species and an eighty-five percent learning curve a shared fraction of three fifths buys 13.63 percent of unit cost. By the threshold $\eta_{\ast}$ under The Omission Is the Penalty, a sharing penalty rising as the square of the shared fraction has an interior optimum only above 15.47 percent, at which point full commonality can be worse than none. By the shortfall $\sigma$ under Why Frequent Refresh Is Worth Something, shortening a design freeze from fifteen years to three takes the capability shortfall from 80.87 percent to 37.82 percent. If the genus concept is worth what the laboratory claims, the value is in the cadence and not in the commonality, and every input to both calculations is unpublished.

One institutional fact, larger than the designation, closes the case. The request procedure requires the requester to research the last approved design number in an official source. The same instruction that states that procedure cancels the publicly accessible list and directs the public to data.af.mil [DAFI 16-401], and that hostname returns no answer from the authoritative nameservers for its own zone. The public source left is one private compiler’s reconstruction obtained under the Freedom of Information Act, which reports the next research number as 69 while the instruction read literally gives 77 [Parsch, DOD 4120.15-L Addendum].

Measured against what the system used to publish, that is a long fall. The founding regulation required the assignment agency to publish, not less frequently than every six months, an unclassified listing of assigned designations carrying the complete designation, the responsible service and a short unclassified description, and the October 1998 list gave three routes to a copy. The 2020 instruction gives one address, and it does not exist [AFR 66-11 and AR 700-26 and BUWEPS 13100.7, 18 September 1962] [DoD 4120.15-L, October 1998] [DAFI 16-401].

So every public claim about the X-67, including every claim in this article, passes through a document the Department did not publish. The designation is empty, the emptiness is irreversible, the mechanism is in the instruction, and the only place a reader can see the gap at all is a table a private citizen rebuilt because the official one was withdrawn.

References

Reference

• AFRL Chooses General Atomics Drone for Off-Board Sensing Station Program, The Defense Post, 6 February 2023, read in full, and SECONDARY, being the source of the February 2023 selection date the laboratory’s own release contradicts
• Air Force Instruction 16-401(I), Army Regulation 70-50 and NAVAIRINST 13100.16, Designating and Naming Defense Military Aerospace Vehicles, 14 April 2005, read in part, and the last edition directing the next available consecutive design number
• Air Force Instruction 16-401, Army Regulation 70-50 and NAVAIRINST 13100.16, Designating and Naming Defense Military Aerospace Vehicles, 16 May 2014, superseding the 14 April 2005 edition, read in part, and the edition that first refuses requests in skipped sequences
• Air Force Joint Instruction 16-401 and NAVAIRINST 8800.3A, Designating and Naming Defense Military Aerospace Vehicles, 9 September 1994, implementing DoD Directive 4120.15 of 2 May 1985, read in full, and the edition in force when the X-49A filled a gap the X-50A had passed
• Air Force Joint Instruction 16-401, Army Regulation 70-50 and NAVAIRINST 8800.3A, Designating and Naming Defense Military Aerospace Vehicles, 1 September 1997, superseding the 9 September 1994 edition, read in part, and the edition in force when the X-49A was allocated
• Air Force Regulation 66-11, Army Regulation 700-26 and Bureau of Naval Weapons Instruction 13100.7, Designating, Redesignating, and Naming Military Aircraft, 18 September 1962, implementing DoD Directive 4505.6 of 6 July 1962 and superseding AR 705-42 of 21 March 1957 and BUWEPSINST 13100.1A of 17 May 1961, signed by LeMay, Decker and Stroop, scanned by Andreas Parsch, READ IN FULL as nineteen page images because the scan carries no text layer
• Castrejon, AFRL’s XQ-67A Makes 1st Successful Flight, Air Force Research Laboratory, 29 February 2024, distributed through the Defense Visual Information Distribution Service, read in full
• Current Designations of U.S. Military Aircraft, compiled by Andreas Parsch, read in part
• Department of the Air Force Instruction 16-401, Designating and Naming Defense Military Aerospace Vehicles, 3 November 2020, sections 4 and 5 and Attachment 2 read directly
• Parsch, DOD 4120.15-L Addendum, MDS Designators Allocated After 19 August 1998, compiled by Andreas Parsch from data obtained under the Freedom of Information Act, read in full as 539 rows and 128 end notes in the copy last updated on 23 August 2026
• Parsch, DOD 4120.15-L Addendum, as archived on 1 February 2026
• Parsch, DOD 4120.15-L Addendum, as archived on 10 December 2025
• Parsch, DOD 4120.15-L Addendum, as archived on 15 January 2026
• DoD 4120.15-L, Model Designation of Military Aerospace Vehicles, 12 May 2004 incorporating Change 1 of 31 August 2018, read in part from a public web archive snapshot after the Department’s own portal returned HTTP 403, and the edition whose Change 1 the article had been told cancelled the public list
• DoD 4120.15-L, Model Designation of Military Aerospace Vehicles, October 1998, reissued under DoD Directive 4120.15 of 2 May 1985 and cancelling the March 1996 edition, signed by Schneiter, scanned by Andreas Parsch, FRONT MATTER READ as page images because the scan carries no text layer
• DoD 4120.15-L, Model Designation of Military Aerospace Vehicles, the last public edition carrying Change 1 of 31 August 2018, at the Department’s own portal, which returns HTTP 403 to every client this article has tried. The document itself was read from a web archive snapshot and is cited separately
• DoD Directive 4120.15, Designating and Naming Military Aerospace Vehicles, 29 November 2004 incorporating Change 2 of 17 September 2020, cancelling the directive of 2 May 1985 and cited by the 2014 instruction as DoD Directive 4120.15E, read in full from a public web archive snapshot of 4 February 2025
• General Atomics Aeronautical Systems, GA-ASI Makes First Flight of XQ-67A OBSS, 29 February 2024, public affairs number AFRL-2024-0708, read in full
• General Atomics Aeronautical Systems, XQ-67A Demonstrates Autonomy and Datalink Interoperability During High Desert Flight Test, 16 July 2025, read in full
• Missing USAF and DOD Aircraft Designations, compiled by Andreas Parsch, the X series and Q series sections and every next-number statement read directly
• Tirpak, Kratos, General Atomics Get Contracts for Off-Board Sensing Station Unmanned Fighter Escort, Air and Space Forces Magazine, 29 October 2021, read in full, and SECONDARY, being the source of the reported contract values the award record contradicts
• USAspending.gov, the federal award reporting system, queried for the Off-Board Sensing Station and Low Cost Attritable Aircraft Platform Sharing programmes

Research

• A Framework on Customer 2021
• A Product Family Design 2008
• Abdelsalam et al 2014
• Abdullah et al 2012
• Abu-Matar and Gomaa 2011
• Achatz 2011
• ade et al 2017
• Adrees 2021
• Advances in Product Family 2014
• Agarwal 2026
• Aghassibake et al 2023
• Agrawal and Cohen 2001
• Aiers et al 1933
• Aircraft Cost Considerations 2010
• Ajibode et al 2025
• Akkermans and van der Horst 2002
• Akoto Chama and Bertram 2018
• Akundi et al 2005
• Albawaneh et al 2025
• Alexandru 2020
• Alexis and Tzes 2012
• AlGeddawy 2014
• AlGeddawy and ElMaraghy 2011
• AlGeddawy and ElMaraghy 2015
• AlGeddawy et al 2017
• Alizon et al 2006
• Alizon et al 2006
• Alizon et al 2006
• Alizon et al 2007
• Alizon et al 2008
• Alizon et al 2009
• Allada and Lan 2002
• Allen 1933
• Allison et al 2006
• Allon et al 2010
• Alsawalqah et al 2013
• Altman 2008
• Amalia et al 2018
• America et al 2000
• Ananno et al 2026
• Andersen
• Anderson 2013
• Anderson and Tushman 1990
• Andrews et al 2020
• Andrews et al 2020
• Aoki and Staeblein 2018
• Appendix 1 Standard nomenclature 2019
• Arakawa et al 2004
• Arellano et al 2015
• Arthur 2008
• Asaga and Nishigaki 2022
• Asan and Polat 2008
• Asher and Maggelet 1980
• Asher and Shafer 1985
• Assef and Geiger 2023
• Athanasiadou et al 2026
• Attia et al 2016
• Aurich et al
• Austin and Rising 2000
• Austin et al 2000
• Ayton et al 1933
• Badiru 2021
• Bagchi and Gutierrez 1992
• Bailey 2008
• Bajpai et al 2025
• Baker 1985
• Baker et al 1986
• Baldwin and Clark 2000
• Baldwin and Clark, The Value of Modularity, Splitting and Substitution, chapter in Design Rules Volume 1, The Power of Modularity, MIT Press, 2000, registry record only
• Baldwin and Clark, What Is Modularity?, chapter in Design Rules Volume 1, The Power of Modularity, MIT Press, 2000, registry record only
• Banas et al 2020
• Bangert and Frances 2017
• Bao et al 2025
• Barber et al 1978
• Barkstrom 2003
• Barriers to Real Options 2009
• Barthélemy and Brucker 2000
• Bateman et al 2007
• Baylis et al 2018
• Bedford 1933
• Bellini et al 2014
• Bellur and Vallieswaran 2006
• Beltramo and Anderson 1977
• Beltramo et al 1979
• Benkoczi et al 2018
• Berber and Yahyapour 2017
• Berenbach
• Berkowitz and Rowlands 2005
• Bernstein et al 2011
• Bibi
• Bick et al 2014
• Bick et al 2015
• Bin et al
• Birrenbach 2000
• Black et al 1990
• Black et al 1990
• Blackburn 2014
• Blackburn et al 2014
• Blank et al 2025
• Blecker and Abdelkafi 2007
• Bloch et al 2023
• Bloch et al 2023
• Boas et al 2012
• Bollapragada and Akella 1993
• Bonanne 2011
• Bonev et al 2015
• Bonilla-Ortiz and Verma 2020
• Borjesson and Hölttä-Otto 2012
• Borjesson and Sellgren 2013
• Bortolini et al 2023
• Boudouh and Grunder 2015
• Boudreau 2007
• Bowman 2006
• Boyce et al 2013
• Boysen and Scholl 2009
• Bright et al 1993
• Brinker 2004
• Brovar and Fortin 2024
• Brown 2004
• Browning 2013
• Broy 2011
• Buckle 1982
• Bulkley and Davis 2013
• Burge et al 2011
• Buschmann and Schwanninger 2011
• Böckle 2005
• Börger and Schulte 2000
• Calvo-Manzano et al 2001
• Cameron and Crawley 2012
• Camm et al 1987
• Caputo and Cardin 2021
• Caputo and Cardin 2021
• Cardin et al 2021
• Cardin et al 2022
• Cardin et al 2023
• Carnduff 2004
• Carrascosa et al 1998
• Carvalho 2023
• Catak et al 2025
• Ceravolo and Bellini 2010
• Chan and Milne 2013
• Chandrakar 2006
• Chandrasekaran et al 2004
• Change Control and Configuration 2008
• Change Control and Configuration 2023
• Change management, configuration management 2018
• Chantzaras 2024
• Chao and Ishii 2004
• Chatras and Giard 2016
• Chatzimichali and Tourassis 2008
• Chen and Barnes 2013
• Chen and Wang 2008
• Chen Chun-Hsien et al 2013
• Chen et al 2009
• Chen et al 2012
• Chen et al 2019
• Chen et al 2023
• Cheng and Zhu 2013
• Cheng et al 2015
• Cheng et al 2019
• Cheung 2002
• Chitsaz and Efatmaneshnik 2023
• Chiu et al 2020
• Choi et al 2026
• Choma Neto et al 2017
• Chong et al 2009
• Choudhuri et al 2019
• Chowdhury and Siddique 2009
• Chowdhury et al 2011
• Chowdhury et al 2012
• Chowdhury et al 2013
• Chowdhury et al 2013
• Chyma 2010
• Clark and Baldwin 2002
• Clements and Weiderman 1998
• Cody and Tiedje 1996
• Cody and Tiedje 1997
• Coleman et al 2015
• Collier 1981
• Collier 1982
• Collier 1984
• Collins et al 1991
• Colombi et al 2017
• Colston 1998
• Configuration Identification
• Configuration Identification 1995
• Configuration Identification 2004
• Configuration Identification 2004
• Configuration Identification for Digital
• Configuration Management and Software 2004
• Configuration Management Identification 2007
• Configuration-Management
• Cook and Coates 2016
• Cooke et al 1933
• Cooke et al 1933
• Corbett and Rosen 2004
• Corliss et al 1974
• Correa et al 2017
• Cosentino 2008
• Cosenzo 2010
• Cost-Estimating Relationships for Aircraft 1972
• Cox 2008
• Creighton et al 1987
• Crompton et al 1900
• Cronie and Thompson 1980
• Crues
• Cui and Weston 2010
• Cummings et al 2018
• Cunha Reis
• Cunningham 1980
• Cunningham 1981
• Curtis 1987
• D’Agostino and Russell 2005
• D’Agostino and Russell 2014
• D’Souza and Simpson 2002
• D’Souza and Simpson∗ 2003
• Dahmus et al 2000
• Dahmus et al 2001
• Dai and Scott 2004
• Dai and Scott 2004
• Dai and Scott 2005
• Dai and Scott* 2005
• Dambietz 2021
• Dang and Nagarur 2000
• Dappert et al 2017
• Darpel et al 2019
• Dart 1992
• Das et al 2017
• Dashofy and van der Hoek 2002
• Davis et al 1986
• de Lange 2002
• De Lit and Delchambre 2003
• De Lit and Delchambre 2003
• De Lit and Delchambre 2003
• De Lit and Delchambre 2003
• De Lit and Delchambre 2003
• de Weck 2006
• de Weck et al 2003
• Decan and Mens 2021
• DeMinico et al
• DeMinico et al 2022
• Department Of Defense Washington Dc 1993
• Department Of Defense Washington Dc 1998
• Desai et al 2001
• Design of Product Platform 2008
• Diamond 1951
• Digital Object Identifier DOI® 2009
• Digital Object Identifier DOI® 2011
• Digital Object Identifier system
• Dillon et al 1998
• DIN 31646 2013-01, Information und
• Ding et al 2026
• Dinnus and Pohl 2005
• Discussion on “The standardisation 1923
• DiVito 1998
• Dolan and Lewis 2008
• Dolan et al 2000
• Domercant 2017
• Donath et al 1981
• Dong and Frank 2006
• Donkels and Voigt 2019
• Dougal and Monti 2011
• Driggs et al 2026
• Du
• Du et al 2002
• Du et al 2012
• DuBois 2022
• DuBois et al 2021
• Dudek and Schulte 2022
• Dunbing Tang et al 2009
• Dunleavy 1988
• Eckert 2006
• Economic models of product 2007
• Eggink
• Eggink and Groll 2020
• Eichstetter et al 2015
• Eilmus and Krause 2013
• ElMaraghy and AlGeddawy 2013
• Emmatty and Sarmah 2012
• Eppinger and Browning 2012
• Ertas and Gulbulak 2021
• Essers
• Ettlie and Penner-Hahn 1994
• Evaluating Product Family Development 2009
• Ewing 2022
• Eynan and Fouque 2005
• Eynan and Rosenblatt 1996
• Eynan and Rosenblatt 2007
• Ezzat et al 2019
• Fabry 1980
• Fan et al 2010
• Fan et al 2024
• Farooq et al 2013
• Farr et al 2024
• Farrell and Simpson 2001
• Farrell and Simpson 2003
• Farrell and Simpson 2008
• Farrell and Simpson 2009
• Feiler and Downey 1990
• Felice 1985
• Fellini et al 2006
• Fellini et al 2006
• Fiala 1983
• Finch 1999
• Fischetti 1974
• Fisher et al 1999
• Flaherty et al 2006
• Flores et al 2025
• Fonder and Headley 2013
• Ford and Dillard 2008
• Ford and Dillard 2009
• Ford et al 2012
• Franke and Hobbs 1993
• Friedrich and Schulte 2025
• Fry 2008
• Fu Li et al 2008
• Fujinaga et al 2007
• Fujita 2002
• Fujita 2013
• Fujita et al 1999
• Gambaryan 2021
• Gansler et al 2010
• Gao and Zhang 2020
• Gao et al 2009
• Garcia et al 1994
• Garmilla Manzano et al 2026
• Garneau et al 2013
• Gaska 2007
• Gaska 2012
• Gasnakis and Homberg 2013
• Gauthier
• Gedell and Johannesson 2012
• Generic product model of 2025
• Gerchak and Henig 1986
• Gerchak and Henig 1989
• Gerchak et al 1988
• Gershenson et al 2003
• Gershenson et al 2004
• Gilbert 1933
• Golovachev 2026
• Gong et al 2017
• Gonzalez-Zugasti and Otto 2000
• Goswami and Allada 2007
• Gould 1974
• Gozzi and Schulte 2026
• Gravina da Rocha et al 2019
• Gray 1979
• Greve and Krause 2018
• Greve et al 2019
• Groocock et al 1933
• Gruber 1992
• Gruber 1994
• Gualberto
• Guangming Chen and McLennan
• Gudmundsson 2014
• Gudmundsson 2022
• Guerrero 1985
• Guinivan 2004
• Gunawan 2012
• Gupta 2021
• Gyllenberg and Koski 1996
• Hallsteinsen et al 2004
• Halman et al 2006
• Hamdi and Tadj 2019
• Hamidreza Kasaei et al 2010
• Hamilton 2002
• Hammarbäck 2022
• Hammarbäck et al 2023
• Han et al 2019
• Han et al 2019
• Han et al 2025
• Hanratty et al 1999
• Hardisty et al 2021
• Harland and Uddin 2014
• Harofteh and Saljooghi 2023
• Hartel et al 1977
• Harvey 1986
• Haspert and Beauregard 1998
• Hata et al
• He et al 2021
• He et al 2024
• Heberlig and Palaoro 1971
• Heese and Swaminathan 2006
• Heikal et al 2019
• Heilemann and Hollatz 2020
• Helmer et al 2010
• Helo 2006
• Henderson and Clark 1990
• Henderson and Higer 2007
• Hendrickson et al 2008
• Heradio-Gil et al 2011
• Herb 2021
• Herbst and Klöckner 2014
• Hernández et al 2015
• Hewlett 2016
• Hewlett 2017
• Hidayat and Fadilla 2025
• Hidayat et al 2017
• Hill 1973
• Hillier 1999
• Hillier 1999
• Hillier 2000
• Ho et al 2017
• Hoek et al 1995
• Hoek et al 1998
• Hoek et al 1998
• Hoek et al 1998
• Hoffman 2010
• Hogan et al 2020
• Hogan et al 2022
• Hohnen et al 2013
• Hoke 2018
• Holmqvist et al 2006
• Holmstedt 2023
• Holt and McNeillis 2007
• Hong and Baek 2025
• Hood 1987
• Horowitz et al 2014
• Hossain et al 2021
• Hou and Lai 2015
• Houshmand and Valilai 2012
• Housley 2013
• Housley 2014
• Housley 2014
• Housley 2021
• How Configuration Identification Works 2004
• Ho¨ltta¨-Otto 2005
• Hu et al 2008
• Huang and Wang 2025
• Huang et al 2008
• Huang et al 2010
• Huang et al 2012
• Huang et al 2012
• Huang et al 2018
• Huang et al 2025
• Hubac 1983
• Hume and Rosen 2012
• Humphrey and Muñoz
• Hwang and Han 2025
• Hwang et al 2022
• Hölttä-Otto et al 2008
• Hölttä-Otto et al 2012
• Identifying Real Options to 2009
• Iman 2025
• IMF Participation in the 2013
• Information and documentation. Digital
• Information technology – Procedures
• Insulation commonality assessment phase 1973
• Introduction to Design Structure 2012
• Islam et al 2022
• Izui et al 2009
• İşci and Günel 2021
• Jacobs 2024
• Jacobson 2004
• Jacobson 2004
• Jakovljevic 2006
• James 1964
• Jans et al 2008
• Jaring and Bosch 2004
• Jelenić 2024
• Jennings et al 2025
• Jeong 2026
• Jeong et al 2026
• Ji et al 2013
• Jiang 2009
• Jiang and Allada 2001
• Jiang and Allada 2005
• Jiang et al 2026
• Jiao
• Jiao 2012
• Jiao and Tseng 2000
• Jiao and Tseng 2000
• Jiao and Zhang 2006
• Jiao et al 2006
• Jiao et al 2007
• Joeris 1997
• Johannesson and Claesson 2005
• John and Villela 2008
• Johnson and Kirchain 2010
• Johnson and Kirchain 2013
• Johnson Lim et al 2011
• Jones 2018
• Jones et al 2021
• Jose and Tollenaere 2005
• Jucevic 1978
• Jung and Simpson 2015
• Jung and Simpson 2016
• Jung and Simpson 2016
• Jung and Simpson 2018
• Jung et al 2017
• Jung et al 2025
• Jönsson and Silver 1991
• Kamrani 2013
• Kane
• Kannan and Nolte 1998
• Karabetsky 2018
• Karagoz et al 2021
• Karimi and Werner 2014
• Kariya et al 2007
• Kashkoush and ElMaraghy 2017
• Kennickell 2016
• Khadke and Gershenson 2008
• Khajavirad and Michalek 2007
• Khajavirad and Michalek 2007
• Khajavirad and Michalek 2008
• Khajavirad et al 2008
• Khajavirad et al 2013
• Khire et al 2008
• Khire et al 2013
• Khreish et al 2005
• Kiam et al 2022
• Kim
• Kim 2026
• Kim and Chhajed 2000
• Kim and Kim 2020
• Kim and Kim 2026
• Kim and Lee 2018
• Kim and Lee 2022
• Kim and Moon 2016
• Kim and Moon 2017
• Kim and Moon 2017
• Kim and Oh 2017
• Kim and Park 2004
• Kim et al 2003
• Kim et al 2010
• Kim et al 2013
• Kim et al 2016
• Kim et al 2019
• Kim et al 2020
• Kim et al 2021
• Kim et al 2025
• Kim et al 2025
• Kima and Chhajed 2001
• Kimita et al 2021
• Kline and Bachman 2006
• Knauber and Thiel 2002
• Kneece et al 2014
• Kohn 2022
• Kokkolaras et al 2006
• Kong et al 2024
• Koo 2013
• Korecki and Hoika 2026
• Korhonen and Mikkonen 2004
• Kosanke and Nell 1999
• Kosanke and Vernadat 1998
• Kota and Sethuraman 1998
• Kota et al 1998
• Kotarski et al 2021
• Kourtis et al 2026
• Kovach et al 2021
• Krause and Eilmus 2011
• Krause and Gebhardt 2023
• Krause and Gebhardt 2023
• Krause et al 2013
• Krishnan 2008
• Krueger 2008
• Krueger 2011
• Krueger 2015
• Kruse and Gries 2011
• Krystallis et al 2026
• Ku 2014
• Kuang and Jiang 2009
• Kumar et al 2004
• Kun and Dong 2006
• Kwak and Kim 2013
• Kwapisz et al 2019
• Kwon et al 2025
• Kässi et al 2008
• Kögel 2008
• Kölmel and Eisenbiegler 2001
• Künzel et al 2025
• Labro 2004
• Lam et al 2020
• Lamine et al 2005
• Lancaster 1998
• Lane 1990
• Lange and Imsdahl 2013
• Larsson and Åslund 2001
• le Maistre 1933
• Learning Curves Production Breaks/Lost 2015
• Learning Curves Production Breaks/Lost 2026
• Lee 2012
• Lee and Jeong 2025
• Lee and Kang 2025
• Lee and Lee 2026
• Lee and Na 2023
• Lee and Park 2024
• Lee et al 2025
• Lee et al 2025
• Lee et al 2025
• Lehnert et al 2019
• Lei
• Lemieux 2014
• Lemieux 2016
• Leuschner 1981
• Levandowski et al 2015
• Lewkowicz 2026
• Li 2010
• Li 2023
• Li and Chen 2008
• Li and Weiss 2011
• Li and Wu 2011
• Li et al 2023
• Li Qian and Zhenyu Kong 2008
• Lim
• Lim et al 2010
• Lin 2004
• Liu 2012
• Liu 2016
• Liu and Hsiao 2005
• Liu et al 2009
• Lu et al 2008
• Lundberg 1956
• Luo et al 2013
• Luo et al 2015
• Lyke et al 2019
• Lyons et al 2012
• Lyons et al 2021
• Ma and Kim 2014
• Ma and Kim 2015
• Ma et al 2017
• Ma et al 2021
• Maccari
• Maccari 2002
• MacDuffie et al 1996
• MacKay 1996
• Maddux 1999
• Maier et al 2024
• Majd and Pindyck 1987
• Majerus et al 1999
• Maldonado et al 2017
• Malhotra et al 2024
• Mandelbaum et al 2008
• Mannion and Savolainen 2014
• Marchese 1963
• Marion et al 2007
• Marion et al 2014
• Martin and Ishii 2000
• Martin and Ishii 2002
• Martinez-Larrosa and Siddique 2026
• Martini and Forcellini 2015
• Martinis 1990
• Masmoudi et al 2017
• Mathur and Jain 2011
• Matinlassi
• Matos et al 2007
• Matthews and Dubois 2021
• Mayall 1983
• Maze 1966
• Mazzola and McCardle 1997
• McClure 1977
• McNeill 1984
• McNicol and Wu 2014
• Mead 2003
• Mehmood et al 2013
• Melber et al 2015
• Messac et al 2000
• Messac et al 2013
• Meyer and Schulte 2020
• Meyer and Schulte 2022
• Meyer et al 2014
• Michelson and Domercant 2019
• Milev et al 2009
• Milne and Chan 2019
• Ming Dong et al 2010
• Minglang et al 2018
• Mirchandani and Mishra 2002
• Missouri Univ-Rolla 1988
• Model-Based Assessment of Variety-Induced 2025
• Model-Driven, Aspect-Oriented Product Line 2009
• Modrak and Bednar 2016
• Modrak and Soltysova 2021
• Modrak et al 2014
• Mohagheghi and Conradi 2004
• Mohebbi and Choobineh 2005
• Mohr 2002
• Monetti et al 2025
• Monetti et al 2026
• Moon and Simpson 2010
• Moon and Simpson 2013
• Moon et al 2009
• Moon et al 2011
• Moon et al 2013
• Morales 2013
• Morales 2013
• Morris and Steiner 2006
• Moskovich 1978
• Motzet et al 2006
• Mukhtar and Agarwal 2009
• Multiple-View Requirements Models for 2009
• Murmann and Frenken 2005
• Murmann and Frenken 2006
• Myrodia et al 2021
• Müller et al 2023
• Nace and Koopman 2001
• Nagarur and Azeem 1999
• Nakao and Iino 2021
• Nanda et al 2005
• Nasrullah et al 2025
• Navarro Gonzalez 2022
• Nayak et al 2000
• Nayak et al 2002
• Nelson 2017
• Nelson et al 1999
• Nelson et al 2000
• Nemet 2006
• Neubauer et al 2022
• Neubauer et al 2023
• Neufeld and Chung 2005
• Newcomb et al 2003
• Nicol and Hooper 1999
• Nidamarthi and Karandikar 2006
• Niemelä 2005
• Ninan and Siddique 2013
• Ninan et al 2024
• Niutanen and Otto 2017
• Nomaguchi et al 2004
• Numerical Taxonomic Methods 1982
• Numerical Taxonomic Methods 2024
• Numerical Taxonomy 1983
• Numerical Taxonomy 2017
• Numerical Taxonomy and Cluster 1994
• Numerical Taxonomy. The Principles 1975
• Nørgaard et al 2025
• O’brien and Colwell 1988
• Ochoa et al 2022
• Oh and Hong 2024
• Oh et al 2022
• Oh et al 2026
• Ohvanainen and Hietikko 2012
• Okpoti et al 2019
• Ong et al 2006
• Opdebeeck et al 2021
• Optimization Techniques for Multi-Core 2025
• Orhan et al 2026
• Origins of Real Options 2009
• Ortiz 1992
• Ostrosi Egon et al 2014
• Otto et al 2013
• Otto et al 2016
• Overview of GBOM Formation 2016
• Ozkaya and Turunc 2025
• Padilla et al 2025
• Paglialonga and Schirnick 2020
• Pakkanen et al 2016
• Pananiswami and Bishop 1991
• Pandey et al 2011
• Pandit and Siddique 2004
• Panter et al 2025
• Park 2021
• Park and Shin 2007
• Park and Simpson 2006
• Park et al 2008
• Park et al 2022
• Parrish 1978
• Pasche and Sköld 2012
• Paskin 1999
• Patria 2013
• Paul et al 2025
• Pedersen and Slepniov 2016
• Pektaş and Pultar 2006
• Peng Chew et al 2006
• Pericon 2024
• Peters et al 2022
• Peterson 1971
• Pfeifer 2019
• PID Service an 2015
• Pierce and Littlefield-Lawwill 2008
• Piligian et al 1968
• Pillai 1956
• Pinckney et al 2023
• Pindyck and Majd 1996
• Ping Jiang et al 2007
• Pinson and Ryan 1972
• Pinto 2019
• Piran et al 2020
• Pirmoradi and Wang 2011
• Pirmoradi et al 2013
• Pisch et al 2024
• Plant 2017
• Pletsch 2021
• Plonsky and Herber 2026
• Pohl and Reuys 2002
• Powers and Schellenger 1989
• Principles of Numerical Taxonomy 2013
• Prinzel et al
• Product Development by Using 2015
• Product Platform and Product 2006
• Programme Management. Configuration Management
• Pudge 1933
• Qi and Xie 2026
• Qian 2011
• Qian 2011
• Qiao et al 2017
• Qiao et al 2018
• Qiu 2010
• Qiu et al 2018
• Quantick 2001
• Quesne 1969
• Quinlan
• Raemaekers et al 2017
• Rahmat et al 2017
• Rai et al
• Rasch and Robert A. 2011
• Ratcliffe 1965
• Raudberget et al 2014
• Real Options Design for 2009
• Real Options in Engineering 2009
• Real Options in Engineering 2009
• Real Options in Manufacturing 2009
• Regan 2001
• Reichenfeld and Jones 1999
• Reinhart et al 2008
• Rendon 2006
• Rendon 2007
• Rendon 2008
• Rennenkampf and Rhyne 2004
• Reynerson
• Reynerson 1999
• Riaz et al 2017
• Ribeiro et al 2007
• Ridder 2024
• Rieker and Haraburda
• Rieker and Haraburda 1989
• Robert
• Roberts et al 2023
• Roger Jiao et al 2007
• Roh et al 2025
• Rojas Arciniegas and Kim 2010
• Rojas Arciniegas and Kim 2011
• Rojas Arciniegas and Kim 2012
• Roncolini and Quaranta 2024
• Rose et al 2014
• Rosen 1996
• Rosen and Choi 2021
• Rosenblum et al 2022
• Roth and Schulte 2020
• Roth and Schulte 2022
• Roth and Schulte 2022
• Roth et al 2020
• Rubio et al 2014
• Ruf and Stütz 2016
• Ruf and Stütz 2018
• Rutenberg 1971
• Rzepa 2018
• Sadraey 2010
• Samarasinghe et al 2019
• Sanaei et al 2015
• Sathe and Pant 2010
• Sauser et al 2009
• Savolainen and Mannion 2012
• Schlee et al 1975
• Schmid and Lee 2013
• Schmitt and Schulte 2015
• Schmitt and Schulte 2015
• Schmitt et al 2015
• Schrage 2023
• Schuh et al 2017
• Schulze and Li 2009
• Schweikhard et al 1998
• Schweikhard et al 2001
• Schwerd and Schulte 2021
• Searle 1977
• Sellon 1900
• Shah et al 2011
• Shamsuzzoha 2011
• Shamsuzzoha and Helo 2009
• Shamsuzzoha and Helo 2010
• Shamsuzzoha and Helo 2012
• Shamsuzzoha et al 2020
• Shamsuzzoha et al 2023
• Shao et al 2012
• Sharan 2019
• Shatnawi et al 2017
• Shaver et al 2016
• Shekar et al 2011
• Sheu and Wacker 1997
• Shi and Zhao 2014
• Shikata et al 2013
• Shively et al 2011
• Shooter 2006
• Shooter and Cohen 2010
• Shoval et al 2016
• Siddapureddy 2011
• Siddique 2006
• Siddique and Adupala 2005
• Siddique and Adupala 2008
• Siddique and Rosen 1999
• Siddique et al 1998
• Simpson 2004
• Simpson 2006
• Simpson 2017
• Simpson and D’Souza 2002
• Simpson and D’Souza 2004
• Simpson et al 1999
• Simpson et al 2001
• Simpson et al 2001
• Simpson et al 2006
• Simpson et al 2010
• Simpson et al 2011
• Singh et al 2002
• Sitcharangsie et al 2021
• Siu
• Sligh et al 2002
• Smit and Pistorius 1998
• Smith 1933
• Smith 2004
• Smith 2006
• Smith and Wandelt 1991
• Smith et al 2003
• Smunt 1999
• Smunt 2011
• Sneath 1985
• Software and systems engineering
• Software Product Line Engineering 2009
• Software Product Line Engineering 2012
• Sokri and Ghanmi 2017
• Song and Zhao 2009
• Sopadang et al 2002
• Sotos et al 2013
• Space And Naval Warfare Center Arlington Va 1996
• Sparrow and Henschke 2023
• Sperling et al 2008
• Srivastava and Rice 2014
• Stahl et al 1984
• Standardisation as applied to 1933
• Staples and Hill
• Sterk and McAtee 2015
• Stettner 2000
• Steuer and Whitcomb 2003
• Stewart and Yan
• Stojanovski and Selva 2017
• Strenzke et al 2011
• Study of the commonality 1975
• Su et al 2010
• Subramanian et al 2013
• Sullivan et al 2001
• Sun and Lau 2019
• Sun et al 2016
• Sundgren 1999
• System Configuration Identification 2005
• Szafranski et al 2014
• Taguchi Integrated Real-Time 2010
• Takai 2018
• Takai 2019
• Takai and Sengupta 2016
• Takai and Sengupta 2017
• Takeda et al 2018
• Tanabe and Awano 2019
• Tang 2008
• Tang 2008
• Tang et al 2008
• Tang et al 2009
• Tang et al 2017
• Tang et al 2017
• Tang et al 2017
• Tang et al 2017
• Tate 1998
• Taylor 2009
• Teacher et al 2023
• Teng et al 2014
• The Modular Operators 2000
• The People’s Liberation Army’s 2025
• Thevenot and Simpson 2004
• Thevenot and Simpson 2006
• Thevenot and Simpson 2006
• Thevenot and Simpson 2006
• Thevenot and Simpson 2007
• Thevenot and Simpson 2007
• Thevenot and Simpson 2009
• Thevenot et al 2005
• Thevenot et al 2007
• Thiel and Hein 2002
• Thomas 1992
• Thompson and Tong 2018
• Thonemann and Brandeau 2000
• Tichy 2001
• Tokunaga and Fujimura 2013
• Tomayko 1986
• Tomayko 1990
• Toppler et al 2025
• Towill 1985
• Transportation Systems Center Cambridge Ma 1989
• Trask and Roman 2008
• Tseng and Jiao 1998
• Tsubone et al 1994
• Tucker 2013
• Turner 2003
• Uddin et al 2018
• Ulrich 1994
• Underwood 2011
• Universal product family design 2012
• Unmanned Aerial Vehicle Design 2024
• Use Modular Architecture
• Using the Learning Curve 2010
• Uusitalo 2014
• Van den Broeke et al 2017
• Van Dyke et al 1990
• van Veenendaal 2023
• Vance 1984
• Varela 2014
• Verhoef et al 2012
• Vines 1989
• Voelter and Visser 2011
• Vogler et al 2026
• Vogt et al 2023
• Voigt et al 2019
• Wagner and Foster 1990
• Wahdan et al 2016
• Wallace et al 2001
• Wallance 2021
• Wallance 2021
• Wallmark
• Wallnau 1992
• Wan 2014
• Wan-Chuan Chan and Arıkan 2020
• Wang
• Wang
• Wang and Tseng 2011
• Wang et al 2008
• Wang et al 2011
• Wang et al 2011
• Wang et al 2012
• Wang et al 2017
• Wang et al 2018
• Wang et al 2023
• Wang et al 2025
• Watza and Cooper 2023
• Wazed et al 2008
• Wazed et al 2010
• Wazed et al 2011
• Wearne 1992
• Wee and Dada 2010
• Wei and Qin
• Weigel et al 2014
• Weingärtner 2002
• Wenlei Zhang and Yushun Fan 2006
• Wettig 2002
• White 1978
• White 2023
• Why do Aircraft Cost
• Why do aircraft cost 1999
• Wijnstra 2002
• Wijnstra 2004
• Windheim 2019
• Windheim 2019
• Windheim 2019
• Windheim 2020
• Wise 2003
• Wong 2013
• Wong et al 2019
• Wright 1933
• Wright 1962
• Wright et al 2022
• Wright et al 2024
• Wu 2023
• Wu and Li 2011
• Wuwer et al 2025
• Xiao 2012
• Xiaogang et al 2006
• Xie and Cardin 2017
• Xu and Jiao 2013
• Xu et al 2020
• Yan and Feng 2011
• Yang et al 2004
• Yang et al 2005
• Yang et al 2013
• Yang et al 2014
• Yang et al 2016
• Yang et al 2024
• Yao et al 2017
• Yao et al 2023
• Ye and Gershenson 2008
• Ye and Gershenson 2009
• Yeager 1987
• Yee Soon Lim 2007
• Yen and Smith 2009
• Yong Lin 2006
• Yongjie et al 2021
• Yoo et al 2021
• Young 1989
• Young et al 2004
• Yuan et al 2025
• Yue Yang and Xibin Wang 2009
• Yun 2019
• Zacharias and Yassine 2007
• Zamani and Beigzadeh 2023
• Zelm and Doumeingts 2013
• Zeng et al 2025
• Zha and Lu 2002
• Zhang 2012
• Zhang 2014
• Zhang and Cheng 2005
• Zhang and Tseng 2007
• Zhang et al 2007
• Zhang et al 2015
• Zhang et al 2019
• Zhang et al 2022
• Zhao 2010
• Zhao et al 2022
• Zheng et al 2021
• Zheng et al 2022
• Zhou and Li 2007
• Zhu 2004
• Zhu 2006
• Ziedorn 2020
• Zimmerman et al 2018
• Zischka and Finnegan 2023
• Zou et al 2026

Related Post