Why Guns Take Years to Get Into Production

Today, let’s take a few minutes to discuss why it’s hard to actually put guns into production. Making a functional prototype is one thing, but the truly hard part is often translating that one prototype into a whole set of tooling and fixtures to mass-produce the design. Generally speaking, the whole development process is a 5-10 year endeavor, and even some of the guns we think of as the most reliable today were plagued with serious manufacturing problems early on.


  1. On a tangent here, regarding the assembly of firearms vice fabrication of parts. Are firearms generally assembled by hand? I don’t imagine robotic assembly lines such as the automotive industry uses, but I could be wrong.

    • Automated assembly is mostly limited to welding and riveting operations. Or for some specific torque sensible assemblies like screwing barrel extension on barrel. This also could serve to reduce risk of work accidents.
      Robots are more common in machining operations, including deburring and polishing. Also in heat and surface treatments and for quality control.

      • Not necessarily true.
        I programmed robotic assembly lines for assembling Anti-lock Braking System modules, diesel injectors, and some other assemblies containing minute parts.
        Yes, a firearm could be assembled by robotics.
        Do they?

    • I’ve seen videos from a Polish AK assembly line and also from large airgun companies. In both cases, assembly was by hand.
      A firearms ease of field stripping makes its assembly a relatively short task.

  2. Thanks for that explanation. I have seen similar problems when developing parachutes. For example, if part “A” needs to fit inside part “B”, tolerances on part “A” might be plus zero and minus .01.” Meanwhile, tolerances on part “B” are pluso.1″ and minus zero. Factories often hear complaints from “the field” when parachutes are packed or operated in ways that never crossed the designer’s mind in his worst drunken nightmare!

  3. Used to do mold work, something like ejector pins would be spec’d at +.0000,-.0001, while the corresponding hole in the mold would be the reverse. And yes, fixturing is critical.

    • I was a member of team with one employer where we were investigating various factors which may be harmful to repeatability of output (mostly with machining). Typically you think of tool wear, but it is also such minute things like presence of tiny chips at fixture’s locating surfaces. Operator blows them out, so he/ she thinks, but they may be transferred from one area to another. When missed they can easily cause a scrapped part.

      • Outstanding points! The field of manufacturing precision includes so many ways to measure to the minutest degree; make machine movements precise, steady, and readily observed / recorded, etc., but no foolproof solution for cleanliness. Also – at least at the level accessible to a hobbyist or small business – monitoring the impacts of minor dings, corrosion, misalignment, or looseness of vises, jigs, and other fixtures is also a mix of diligence, feel, and luck.

          • One thing that’s pretty close to guaranteed is your educational comments. I’ve learned more from your remarks on this article alone than from certain whole engineering courses at school.

  4. This is a high quality lesson by Ian. He put a finger on it in understandable layman’s terms. Since I worked in firearms engineering many years ago I know what is involved. I could not say it any better.

    I recall numerous tolerance (clearances are not tolerances but are closely related) studies back and forth to verify relations of parts in various stages of wear/ contamination in addition to influence of hardness and protective finishes. Question often being asked is: “what part do you want to wear faster without effect to reliability”; “what is cheaper and easier to replace”; “what should be in a replacement parts set”? What is life expectancy of weapon as a unit (typically most wear takes place due to training)? How does it project into cost for customer? Can we still make a profit? A pile of questions with few solid answers.

    Then you have behind your back manufacturing people who want tolerance to be “as wide as possible”. Every decision has to be backed by test and documented. Believe me, it is “just mechanical” but still a substantial field of study.

  5. Every time I have the opportunity to see manufacturing drawings of gun parts I like to look at tolerances in critical areas. Let’s be clear here – no firearm, or more specifically military rifle has “loose tolerances”, leave along “suitable for less developed countries”. That’s nonsense, just forget it.

    It is just a matter how tight they are and mutual relationship of features (such as parallelism, concentricity or squareness). I became familiar with AR bolt group components and I know they are indeed demanding in that respect. So are AK bolt components, but not as much. Mainly, they are geometrically not as complex and puny.

    Now, how manufacturing guys deal with it? They are the ones who have to make a “do” out of it. Did you say by using CNC? You are right, but consider all the influences during the shift, mainly tools/cutters wear, effect of vibration and so on. How you setup tool bits in such a way that is cuts a tiny bit more at beginning and tiny bit less at the end of shift, to be about “right on” in the middle of it? Tool path is assumed the same, as set into CNC program.

    By so called “offsets”. Those are intentional values entered into tool/ tool holder setup assembly. Their values are obtained by previous statistical process analysis. If you manage to get your production thru the 8 hours shift which falls into drawing tolerance zone you are – a King. Zero scrap is the objective. Then you check them accordingly to previously established QC plan, sign them off and ship them for heat treat and protective finish. Done, get paid and go home. Next day same thing πŸ™‚

  6. I recall an anecdote, I believe from Peter Laidler’s Sten book, that the reject rate for brand new MKIII Sten’s was so high that a system of gauging for parts before installation was proposed.

    They then realized that simply adding the time to check if parts were in spec would reduce production by some 400k guns per year, and they weren’t discarding that many new guns as nonfunctional, so they simply carried on. To me THAT has always been the best example of a design that is optimized for maximum efficiency, when it is better to throw away the odd bad item than take the time to do it perfectly. Better can at times be the enemy of ‘good enough.’

  7. And…everything didn’t switch to 100% interchangeability just because of Eli Whitney. Recalling the vids about the Reising submachine gun.

    • Absolutely. Just as one example, interchangeable barrels with pre-headspaced extensions didn’t become common until WWII, and are still far from universal. Even today there are manufacturers congratulating themselves because they provide a tool and/or trick to help the shooter through a process that should have been “set it and forget it” at the factory.

  8. Interchangeability is another good one and directly related to mfg. methods.

    I have seen production at beginning of 1980s which was aimed to be fully interchangeable. It can be done but it needs corresponding manufacturing technology/ methods. The previous generation (such as FAL) was meant to be interchangeable but not fully. For example the Locking shoulders were inserted into receiver while the locking distance was individually measured. After insertion of correct one, they were staked in place.

    That was pretty common in older days. Actually, selective assembly is not bad, if user can live with it during the live cycle; it is cheaper for manufacture (more open tolerances). You just have to select parts into groups, which is bit more involved in terms of handling/ labor time.

    In time of my military service (with vz.58), were were discouraged to swap parts from rifle to rifle although I suspect they were interchangeable. There was basically just one serial number. The most sensitive was considered gas cylinder bore diameter. We were prohibited to use any sort of abrasive material during cleaning. No wonder, it was not adjustable.

  9. It’s about time someone explain that there is no such thing as a perfect fire right off. After reading hatchers book on M on grand I understand it fully. I’m glad you took the time to explain it

  10. Geometric dimensioning and tolerancing is what manufacturing went to. It involves things like minimum material condition, uses set datums where dimensions start, and has tolerances on things like how circular a hole is, or how parallel something is. Does not look at all like the blue prints in a do-it-yourself magazine. There were hints of it during WWII, but was formalized afterwards.

    For a long time, part of making good parts with mass-production has been having good gages to test them with. Especially how things work in three dimensions. Ingenious designers and skilled toolmakers are the ones who make them, sometimes complex shapes with moving parts that are held to ten-thousandths of an inch. These are often at each machine tool, and operators check parts coming off the machines, certainly at the end of each batch, to see if things are still in spec. I that that Garand’s skill as a tool maker is what saved the M1 project. Only someone that dedicated and with those skills would have figured out how to actually make the thing.

    The effect of tolerances going together is called “tolerance stack.” Something the Japanese do right (starting a few decades ago), I think, is to not see upper and lower tolerances as the line to cross and then move on to other things, but rather they use tools like design of experiments to economically get as close to the target as possible. Ford and Mazda once each made the same vehicle to the same tolerances, but the Mazda was still running smooth after the Ford started rattling.

    Selective assembly (i.e., not full interchangeability of parts) is probably still around. It certainly was when I worked in the car parts industry some years ago. After some manufacturing operations, it was just more economical to have something automatically gaged then have a light indicate which of a few parts of slightly different size to use to complete the assembly. That was with drive-line components. Used to at least, pistons were machined to different sizes and selectively installed into cylinders for car engines.

  11. And here I thought you had to spend the time to negotiate and train the magical gun gnomes

    Consider me enlightened

  12. I’ve said the same thing on here more than a few times in the past. Serial manufacture of precision parts is hard.

    Every one of these guys who says “Oh, yeah… If only I had a time machine to take an AK47 back to 1860 to hand off to Party X, that’d be super-cool…”. Nope. Not gonna help, and would probably lead Party X down a rathole of attempted copying that would bankrupt them and leave them in worse shape than historically. Harry Turtledove, I’m looking at you!

    You could probably manage some effective change by making earlier introduction of next-generation tech like passing off the Minie ball to Napoleonic War-era forces, but the wholesale introduction of things like the AK to even the WWI era would create huge manufacturing issues. Dropping back ten-twenty years like dropping the AK on 1930s Britain or America? Yeah, in select cases, that’d work–So long as you could convince them to take up the intellectual framework necessary, right along with it. You’d still have to allow for then-current production technology, as I’m fairly certain that trying to produce the M16 in the 1930s industrial milieu might end disastrously. The concept of the assault rifle would probably work, but the more sophisticated manufacturing requirements for something like the forged aluminum receivers would likely run up against the need for that technology to be restricted to aviation use alone…

    All in all, it ain’t as easy as the various and sundry science fiction authors would have it. Very few people are at all cognizant about what goes into manufacture, at all.

    • Hi Kirk

      I see the M16 architecture as (excuse me, no offence intended) a Landing gear strut made to shoot bullets. Its origin, the technology it exploits, is precisely that. IF in 1930s someone took a piece of aluminum forging (which they knew how to do) and scaled down corresponding cartridge, they could have had it. Four lug bolt instead of 7 with one phantom one, does not make that much of difference. No revolution there. Definitely not Golden calf to bow before.

      We all know, that in firearms the progress in slow, painfully slow. Two steps forward, one step back. Why, if all other technologies are moving in leaps forward?

      • Denny… In the 1930s, the landing gear strut you call out was not a routine bit of manufacture. Instead, it was cutting-edge in metallurgy and processing–Do note the rather large amount of problems with that “everyday” technology well into WWII, with all the various and sundry issues with collapsing landing gear.

        You could probably make the things, sure–In small quantities with huge rejection and failure rates. Large-scale manufacture, churning out a few million for the war? Probably not, and if you tried, you’d likely eat up capacity to manufacture things they really needed–Like, aircraft landing gear struts.

        Manufacture is the art of the possible, and I’m afraid that it just wasn’t as possible as a lot of hand-waving authors of fiction would have it.

        Now, sure… The AR-18? Likely quite doable, if you could produce the necessary powders and other supporting things. The AR-10 and derivatives? I am pretty sure that the state of the art would have allowed for production, but likely not mass production.

        You could argue this for hours, days, and weeks. Actual answers would have to come from someone who knew the ins and outs of the involved technological histories, and those folks are vanishingly rare. One I know and discussed this with was dismissive of the likelihood of it all, and what I’m saying is based on what he told me about it. Most of the AR-10 was deliberately designed on the things they’d learned in WWII about precision aircraft manufacture, and it was intended by design to exploit those things. Mass production before that capacity existed probably could have been achieved, but you’d have had to invest what they did into the aviation industry during WWII in order to make it happen.

        • Kirk, I was referring to new generation of mid-size all-metal planes like DC-3. They represented all what was out there at the time. But, let’s leave it as an open argument without clear conclusion. It is not that important anyway. Fact remains that the AR-15 is based on aircraft landing gear type of components, forged and machined. To me, that was a major progress – light and strong with inline layout.

      • Now, pertaining the point I am making about slow and tedious development in firearms, I want to bring couple of post M16 examples with question IF they meant real progress or not. How much time it took to bring them into service and with what gain? I realize this is a bit over the envelope Ian had on mind; he meant more as an individual type/ model.
        – Steyr AUG, progress – yes or no
        – FNC, progress – yes or no
        – HK G36, progress – yes or no
        – Beretta ARX160, progress – yes or no
        – Kalashnikov AK12, progress – yes or no

        I have purposely chosen types in same or similar caliber. In my view they are not necessarily a progress in a major technical-tactical aspects, part of ability of mentioned types to mount full folding stocks.

        Is it the material for major structure components which is synthetic material? But their empty weight is no less than 3kg. In comparison, the sa. vz.58 weighs 2.7kg with steel billet receiver. So, where is the “progress” during last 50 years? They still kick albeit little less than rifles with full power ammo.

        • “….part of ability of mentioned types to mount full folding stocks.”

          I wrote this in early hours and missed to add:
          …. part of ability of mentioned types to mount full folding stocks, including optical and accessory rails.

        • The real progress represented by the AUG is in my view primarily the use of an optical sight on a general issue standard rifle (1977).
          I admit, the advantages of its bull-pup layout I only recognized after actually shooting it.
          Having carried the G3 in my military service, which was not really heavy in my view, I fail to understand the popular whining over 3 kg versus 2.7 kg.

      • “(…)in firearms the progress in slow, painfully slow. Two steps forward, one step back. Why, if all other technologies are moving in leaps forward?”
        That hints that there was not disruptive innovation of great magnitute or its effects did not unfold yet, for few decades. Nothing which would push prior existing designs into obsolete.

    • “(…)AK to even the WWI era(…)”
      This reminded me about Cei-Rigotti carbine
      this was more-or-less automaton but created in 1890s and proved too advanced
      While many militaries were testing semi-automatic rifles before the WW1, very few armies were ready to actually adopt any of those advanced, complicated and expensive designs. As a result, the Cei-Rigotti rifle never proceeded past prototype stage, with only few specimens made in rifle or carbine configurations, semi- or full automatic.

  13. Excellent video, thanks Ian! There is also the business side, as you allude to, above and beyond figuring out profitability: having to keep shareholders, investment partners or government procurement people on board as you go through this long production design process and (hopefully) make it work.

  14. The subject of tolerance studies has always been a good one (partly because they are complicated and nobody likes doing them). Someone brought up the need to anticipate component wear over time, which generally tends to result in “looseness” within a mechanism. In the weapons my group used to develop, the opposite situation was also a necessary part of our design paradigm – fouling, corrosion and wear that might result in the displacement of material instead of its removal.

    Our tolerance studies began with the (relatively) simple problem of complete part interchangeability of the manufactured parts (any pile of the correct, dimensionally-compliant parts could be assembled and result in a fully functional weapon), but then we also had to look at the impact of dumping fine Middle Eastern sand into the mechanism without jamming it up. A gun that worked perfectly in the development lab may fail miserably in sand, dust, fungus, ice, you-name-it environment because those variables weren’t also considered during the design tolerance studies. Likewise, we initially design it to fire the first round… but we also needed to consider what the weapon looked like when it fired its 500,000th round (which also influenced our initial estimates of maintenance requirements and part replacement intervals).

Leave a Reply

Your email address will not be published.