(Note: there’s an update and acknowledgements at the end, which you might not have seen if you got this by RSS or read it between 0600 and 0830R 9 Nov 13).

Afghans have a saying about the cruel and charismatic guerrilla leader Gulbuddin Hekmatayar: “If a man is ever once Gulbuddin’s man, he’s Gulbuddin’s man for life.” That saying reflects the fanatical loyalty Gulbuddin’s minions feel towards him.

The equivalent in the gun world is Heckler & Koch fanboys.

There are good and logical reasons to be an HK fanboy. The guns are well-engineered, and better-built; they’re typical products of German engineering, in that they are both complex and elegantly designed. The HK light machine guns have never had the commercial success of the rifles and submachineguns built on the same principles. But this is not because they’re badly designed; HK weapons, when they are not the best in the world, are usually among the best in the world.

The HK23E below is a 5.56mm light machine gun descended from the gun we’re talking about today: the XM262 SAW.


The XM 262 came from behind to nearly steal the SAW competition from the two preferred candidates, the XM 248 and XM249. And yet, it wasn’t a purpose-designed machine gun at all, but a very clever modification of an ordinary service rifle.

The deep back story

We’re going to breeze through the early history of H&K and the roller-delayed system because we assume a basic familiarity with them (perhaps longtime commenter Oberndorf will have something to add). If you want the true details, particularly the technical ones, the reference you need is Full Circle by R. Blake Stevens. In essence, H&K was formed by former Mauser engineers to do general industrial machine and manufacturing work during the immediate postwar period. (Both Mauser and firearms work in general were shut down during the early occupation period, and it was not obvious that Germany would evolve into two sovereign states within a few years, with only Berlin remaining under de jure 4-power occupation).  In that same era, another group of Mauser veterans had relocated to Spain to continue work on a roller-locked firearm originally developed by Karl Maier and Ludwig Vorgrimmler at Mauser under the name MP45. The MP45 was a 7.92 x 33mm assault rifle which used the round and magazine of the successful MP44 but introduced an ingenious new locking mechanism.

There are at least two explanations that purport to account for the engineers’ presence in sunny Spain, and we make no judgment as to their validity. The first was that Spain wanted firearms engineers, a field which seemed constrained in occupied Germany, nearly deindustrialized by bombing and ad hoc “reparations” by some of the occupying Powers. The second is that members of the occupying armies wanted to have a talk with all former Mauser engineers and managers about the firm’s utilization of slave labor during the war.

While in the cozy embrace of Franco’s Spain, the engineers developed a delayed-blowback rifle, originally for intermediate cartridges but then adapted to a reduced-power NATO round. This was the CETME, named for the acronym for the place they worked: Centro de Estudios Técnicos de Materiales Especiales, a Spanish RDT&E think-tank, which would become the standard Spanish battle rifle. The gun’s leading developer was Ludwig Vorgrimmler, who returned to Oberndorf am Neckar with his team after putting Spain on the rifle-design map for the first time in the nation’s long history (Spain continued development of indigenous small arms at CETME after the Germans’ departure).

Vorgrimmler and his team found employment at H&K, which was then making sewing machine parts. The CETME design was well adapted to modern (circa 1955) production methods; its receiver was a weldment made of stampings (“pressings” in British) and many of the smaller parts could be produced by analog-automated machinery like centerless grinders and screw machines. H&K continued to develop the design into the HK51 and it was sold in many versions, becoming the national service rifle of a swath of Europe from Norway to Greece and Turkey, including Germany, which knew it as the G3 (it was the third rifle — Gewehr — adopted by the Bundeswehr, and the first German-designed one). The G3 led German arms exports to Central and South America, where it was hailed as the successor to well-loved Mauser 98 variants in many places. A semi-auto version was even sold in the US until the Clinton ban, as the HK91. (These have become high-priced collector items since the truncation of supply in 1994, and have spawned American clones of varying quality).

H&K wasn’t making sewing machine parts any more.

The H&K roller-locking system is unique. (For details, try US Patent 3,283,435 from 1966, the US instantiation of a German patent, and for an explanation of the physics and mechanics involved, this transcribed 1970 H&K document at HKPro can’t be beat). While the Rheinmettal-Borsig MG42 had a short-recoil roller-locking mechanism (based on a prewar Polish patent), it was fully locked at the moment of firing.

HK roller lock force diagram - techfig5The lighter, simpler, more easily manufactured H&K system was not fully locked and its principal benefit, then, was that the barrel did not have to move before the mechanism began to unlock, and no gas needed to be tapped off and managed to force this unlocking. Additional benefits included “automatic” adaptation to a wide range of ammunition compared to gas- or recoil-operated guns. In fact, the whole H&K list of advantages for their system (from the 1970 document mentioned above) includes:

  1. The advantages of the simple inertia bolt are retained, in particular the fact that only the pre-determined distribution of momentum and the area of the gas force curve throughout the time are of importance for the recoil velocity of the bolt.  This provides very good adaptability to all types of ammunition of the same caliber, bullet weight and velocity without the same adjusting elements in order to compensate for the shape of the gas force curve.
  2. The low extraction velocity most practical for the cartridge case is ensured by the physical and design principles, providing extremely secure cartridge support.
  3. The movement sequence of the bolt assembly and the receiver follows the gas pressure sequence without any delay.  The uniform, play-free commencement of all movements of the bolt parts and receiver avoids sudden, uncontrollable impacts.
  4. The bolt does not make any rotating or tilting movements when opening or closing.
  5. The bolt parts are arranged symmetrically to the axis of the bore.  The roller contact points are only at a small lateral distance from the axis of the bore.
  6.  The sequence of the reaction force is uniform and without distinct peaks of force.
  7. Because the cartridge case pushes and does not have to be pulled, extractor strain is limited to ejection.

Points 3 – 6 contribute to the extremely high accuracy of the rifle.

A cutaway H&K chamber, showing the flutes  and the chrome chamber (the hole is not a factory feature). This barrel is on GunBroker at present.

They don’t mention the relative ease of manufacture of such a system, which was a major consideration in the design. Some points might be disputed. For example, “low extraction velocity” (point 2) or not, early prototypes were bedeviled by extraction problems, caused by adhesion, which were resolved by fluting the chamber and plating the bore with hard chrome.

The mechanical disadvantage built into the system, and the difficulty of forcing a change in direction of the parts, was enough that this much lighter mechanism was just as safe with a powerful round as the massive MG3 version of the MG42 was with the same round (for grins, H&K calculated the weight a straight-blowback 7.62×51 bolt would have to have: 14kg/31 lb.). Previous delayed-blowback systems such as the Czech Koucky’ system used a sliding steel mass to provide the inertia; the later FAMAS would use a lever. H&K cammed tungsten rollers (tungsten for its heavy weight) into a machined groove in the receiver, where they didn’t exactly lock in the sense that the MG42’s rollers did, but required quite a bit of force to make change directions.

From rifle to machine gun

The success of the G3 spawned a wide range of variants, some of which varied by caliber. 5.56 and 9mm versions were common, and 7.62 x 39mm versions were experimentally made — but didn’t sell. Some of the designs varied by purpose, too. The highly modular design was readily adapted to a sniper rifle, or to a short-stocked paratroop or CQB weapon, or downsized to make a pistol-caliber submachine gun or tiny personal defense weapon. Making a machine gun seemed possible and logical, so H&K did.

In fact, the gnomes of Oberndorf made not one, but two, machine gun designs based on the G3, and later versions would be modularly interchangeable one with the other. The HK11 used the magazine of the G3 (or a rare 30-round version, and still rarer drum) and the HK21 used a belt feed. Both took the basic G3 and added a clever barrel-change mechanism, something air-cooled machine guns need to manage heat. (It’s most important on guns that fire from the closed bolt, as the HKs, with their rifle-derived trigger mechanism, did). Too much heat buildup damages barrels and is inimical to accuracy, but it’s also unsafe, as rounds can cook off in a hot chamber after a cease-fire. (In a really hot chamber, rounds can cook off out of battery… bad, bad, bad). So LMGs are normally provided with a spare barrel and some ability for the crew to change the barrel. With two barrels, alternating, a crew can sustain a higher long-term rate of fire than they can nursing a single tube.

Barrel change on the HK21, etc., couldn’t be simpler. Grab the insulated barrel handle, rotate, pull. Don’t touch the metal bits (you will only do this once, if you ever do it). Take the replacement barrel, push, rotate, click. Back on target.

The guns have an integral bipod and can attach to a pretty sophisticated tripod. After the HK11/21, the company redid the whole thing as the HK13/23 in 5.56mm. These guns used the smaller receiver dimensions and parts of the 5.56mm rifle.

XM262011nvyThe belt feed is different from any other machine gun in several ways. The first is that, unlike Kalashnikov going from AK to PK, or Saive and Vervier going from BAR to MAG, the wily Teutons didn’t flip the mechanism to put the belt feed on the bottom. (This was a little like the Rodman Labs crew had done, although they had their belt come up and over internally under the influence of a ratchet). This was not what gunners were used to; all the way back to World War I, feed trays were on top. It wasn’t necessarily a bad thing: an HK21/23 gunner can set a new belt in without skylining himself, compared to his buddy on an M249 (Minimi) or M240 (MAG).

There were several different iterations of the belt feed. Initially, the HK21 used the German issue DM1 belt, a nondisintegrating belt. It has a sprocket drive, although one more complex than the XM248. The belt feed is simpler and easier to disassemble than the usual MG42-style feed, but it wasn’t designed to be readily disassembled: many of the pins were roll pins, which are ill-suited to that kind of maintenance.

Coming to America

XM262_012yvH&K tried hard in the 1960s to break into the US market. Its 5.56 rifle was tested by SF and SEALs in Vietnam, who liked its available 40-round magazine at a time when M16s and CAR-15s let you seek fire superiority in 20-round increments. For a time, H&K’s guns were represented to the military by hard-luck Massachusetts manufacturer Harrington and Richardson, which always had some good reason when they failed to deliver contract guns or parts on time or at quality, and only stayed in the game because of the seniority of Massachusetts politicians.

The SAW competition which engaged the Army for most of the 1970s seemed like it was made to give HK an opportunity to get into the Free World’s largest arsenal. The Army wanted a belt-fed, 5.56mm machine gun, by God the team in Oberndorf had it ready to go. So H&K brought its HK23 to the test, but the Army had settled on a 6.0 x 45mm caliber. H&K was leery of a speculative redesign for a cartridge with limited support, but when the Army wanted some guns for comparison purposes, they provided them.

At this point one of the most controversial things occurred: the HK23 failed on reliability grounds.  H&K’s executives strenuously objected to this result, and blamed it on the Army’s substitution of ammunition. They had to lobby their way back into the competition, alongside their usual competitors from FN and the unusual but promising Army-Ford XM248.

The XM262 in the 1978-79 DT/OT

The gun provided for the XM262 trials was not the same one tested earlier. It was a 5.56 gun built on the larger frame of the HK21. The reason for this had less to do with the Army test and more to do with the fact that H&K had found a new religion: modularity. With a pop of a couple pins and a swap of a feed group, the gun could be converted from belt to magazine fed. Swap a bolt group as well, and you can change calibers.

Like the other makers, H&K was tasked to provide 18 guns, and like FN, H&K was given no contract to produce the new gun. The H&K guns were all hand-tooled prototypes, and they were all labeled: PROPERTY OF HECKLER AND KOCH. They resembled a G3 rifle, but with what we now know as HK21 features: interchangeable barrel, adaptability to NATO standard tripods and optics. They also provided two belt feeding mechanisms and bolt groups, one for each caliber.

The belt mechanism was the subject of considerable tinkering. The 5.56 version always used individual links, and the 7.62 version was made, at one time or another, in versions that took DM1 belts, American links — and both. (The next picture is not an H&K, but a Michael’s Machines clone of the later, shorter-barreled (~17″) HK23E — in x-rays).


The XM262 was unique among the entrants in that it fired from a closed-bolt. The open bolt used by the other MGs is a trade-off which gives up some accuracy for rate of fire and safety. Open bolt guns don’t have sniper-rifle accuracy, and they also don’t risk cook-offs by letting a chambered round sit for a long time before firing. The H&K gun did approach sniper-rifle accuracy, and it had a rear sight designed to give the gunner a shot (no pun intended) at exploiting that accuracy.

The good news is that the reliability performance of the XM262 in the trials was much better this time. The bad news was that while the Army was fascinated by the easy conversion from 5.56 to 7.62, it wasn’t a feature their end users had asked for. (Probably because the end users were unaware it was possible).

To win the competition, the XM262 would have to edge out the Army’s own baby (XM248) and its fellow-European counterpart, the FN XM249 Minimi.

Hindsight lets us read in the certain knowledge of the Army’s decisions, but in early 1979, the executives of H&K must have been feeling cautiously optimistic about their Squad Automatic Weapon entry.

When this series resumes next week, we’ll look at the last competitor, the XM249 Minimi, and wrap up what we’ve learned about that competition, over 30 years ago.

UPDATE 0830R 9 Nov 13

This post initially went live without necessary acknowledgements: we’d like to thank Michael Otte of Michael’s Machines, who is as humble as he is expert, for patiently answering our questions. Any remaining errors in our description of the HK21/23 and XM262 are ours alone, and suggest we did not absorb his tutelage adequately. Now we want to buy one of his MM23Es. You should, too.

It occurs to us that we should have mentioned the fate of the test guns. The XM248s belonged to the Army, as they were built under Army contract. With only a half-million-dollar budget in FY79, the Army couldn’t buy 18 guns each from FN and H&K, so they asked the manufacturers to lend the guns. Afterwards, the guns were returned to the companies’ US representative. Only H&K and the ATF (which under law, cannot say) know for sure what happened to the XM262s after that, but it’s probable that most of them were deregistered and scrappped. Inventory costs money, especially inventory produced bespoke for a customer that didn’t accept it. At least one gun, Operational Test Weapon No. 12, remains in H&K possession today (we’ll have to check and see if the one in the Gray Book is the same, or another example accounted for. It may shock you but coffee-table gun books don’t all travel with us). It’s Serial Number A037812, as can be seen in the photos here, which brings us to our second acknowledgement: those photos came from this thread at HKPro, where a member posted photos of an H&K display that also shows off a rare take-down PSG2. (If you don’t know what that is, it’s an extensively modified, sniper version of the G3/HK91. If you do know what it is, or want to know what it is, you need to be a member of that forum).

This entry was posted in Foreign and Enemy Weapons, Industry, Machine Guns on by Hognose.

About Hognose

Former Special Forces 11B2S, later 18B, weapons man. (Also served in intelligence and operations jobs in SF).

15 thoughts on “The SAWs that never WAS: Part 4, H&K XM262


Great series of articles. Minor point – by 1945 the ‘MP’ designation had already given way to ‘STG’, so the new roller-delayed blowback rifle was designated STG 45(M) when trials rifles were ordered.

Hognose Post author

Quite right, Jonathan.


Thanks very much for the information! This post and the links provided will occupy take some time to read through and understand I’m sure.

On a slightly different topic, some time ago while surfing for information an article was found on ways to enhance the M-16/AR-15 platform. I’m sorry to say I do not remember the author or date, but do remember is was quite an interesting article. By chance could you find time to expand upon this topic (amongst others you probably have in consideration)?

The direction taken was to look for, what seemed to be best in class materials such as barrel steel (4150 cold hammer forged w/ chrome lined), BCG finishes/coatings (NiB), etc. Considering the state of affairs after “someone” gave a speech some items were notably difficult to source. I would be particularly interested to see your thoughts and suggestions for other surface treatments such as Melonite barrels & gas tubes, or enhancement recommendations for increased reliability and service life.

Along the sames lines it would be interesting to see your view on varying buffer springs/buffer weights for varying 5.56 cartridges (193, 855, etc) and how this corresponds with carbine or mid-length gas systems.

Thanks and best regards,

Hognose Post author

Boy, that’s a mighty open-ended task there: “improve the AR.” The DOD and all its contractors, foreign firms like FN and H&K, and even the Chinese have been trying to do that since cars had tailfins. There are literally tens of millions of words of argument, opinion, and solid signal amid the heavy noise of popular forums like AR15.com and M4Carbine.net.

Funny thing is this: “improvement,” once you’ve decided what direction you’re improving in, is an asymptote. Like a 1911 or a Glock, an AR from most reputable manufacturers is a good general-purpose gun out of the box. For instance, in my last three or four military M4s (all made by Colt) I literally never had a jam that was not due to a bad magazine. Several of my retro AR clones have never jammed (none of them gets the wringing out my service guns used to get). Most of them are more accurate than a military rifle needs to be.

So, where do you improve? Accuracy sends you one way. Harsh-conditions, low-maintenance reliability another. Holding up to high rates of fire in a third direction entirely. What about corrosion resistance? (That’s what drives a lot of surface finishes, and why military weapons have had chromed bores for as long as 70 years now). Finally, there is a downside to customization, in that customization is personalization. That’s great for you (and may be emotionally satisfying for your heirs). But what about when you want to trade the 2013 state of the art for the 2023 state of the art? You won’t recover the money spent on personalization, and you may have difficulty moving stuff that was popular now but turned out to be a fad. (I came *this close* [Maxwell Smart finger gesture] to buying an early 6mm CAR-15 wildcat from an outfit called Old Sarge in Lytle, TX, because “it was the wave of the future” in 1980).

Still, a great idea, and also there might be some room to talk about how to set up an AR for specific purposes.


And that’s why I like coming here. There are always differences in approaches and consideration (sometimes limited by not having extensive familiarity with noisy devices, their background development, or lack of two-way range time). Working from those limitations on my end a good reality check is always welcome.

As noted, the personal customization might be desired by a single user only. The more input and background that can be accumulated helps to sort out the wheat from the chaff. Thanks for the additional perspective.

Samuel Suggs

am I the only one who thinks the cable lock on the feed mechanizm is a little absurd looking?

Hognose Post author

Some shows require some stuff like that. ISTR that SHOT makes you take out all firing pins.

Samuel Suggs

Ah, makes sense, I thought it was a factory exhibition photo or a collection photo never occurred to me that it might be an industry or public gun show.


This is an excellent series. I’m definitely looking forward to the next installments.

A nitpick; I don’t think that the rollers in an HK rifle/MG/SMG are tungsten. My copy of Full Circle is on loan to a friend, but if I recall, those rollers are steel that have bee surface nitrocarburized to 65 Rockwell C. A few of the precision rifles use titanium rollers for some reason. Perhaps you’re thinking of the granulated tungsten buffer that reduced felt recoil on later-model G3s and most subsequent HK long guns? It’s not the mass of the rollers that matters; it’s the mass of bolt carrier and the mechanical ratio of the initial movement of the bolt carrier in relation to the bolt. It’s basically the same system as the Kiraly lever blowback in the FAMAS, except instead of using a lever to generate leverage they use two wheels and four inclined planes. I’m sure it makes sense to Germans.

Hognose Post author

I’ll check the tungsten ref out and fix it if I am wrong. My copy of Full Circle is at home and I am not.

The reason for using the rollers instead of the lever, because just as you did, the 1970 HK doc notes that it’s basically just a lever laid out differently, is to keep as much of the action as possible inline and to reduce anything that would impart significant translational or torsional force to the firearm.


I was being flip; the roller system is actually probably easier to manufacture. For one thing, you don’t have to mill out this tricky looking bugger:


The FAMAS accelerator lever looks a bit tricky to do in the days before ubiquitous CNC machining centers. Certainly it looks harder than the centerless grinding HK could use to make their rollers.

The lever system has fewer parts, but they are much more complex shapes. The roller system has more parts, but they are extremely simple cylinders and wedges.


I’m not too sure that I agree with the statement made about “relative ease of manufacture”, at all. There’s a reason HK hasn’t manufactured or designed a single roller-locked weapon design in the last not-so-few years, and the reason for that goes right to that whole “ease of manufacture” issue.

A roller-locking system has to be carefully designed in regards to a specific ammunition loading, and when you start sticking a wide variety of loads into one, you start seeing issues. The cartridge also has to be suited to the design, as well-The old 7.92X33 Kurz was what they were initially designed for, and while it was adaptable to the 7.62mm NATO loading, the adaptation to the SCHV 5.56mm format was not as easy. Many roller-locking designs for that caliber have foundered on the issues of ammunition loads, and ease of manufacture.

The Thais are rumored to have given up on the HK33 series due to the uneconomical rejection rate they experienced on the production line they bought from HK, once they quit paying HK for tech support. The Spanish never really got the CETME Modelo “L” to work properly, outside of the hand-produced prototypes. The problems with that design were so large that the notably economical Spanish military replaced the Modelo “L” within 15 years, going to the G36.

I’ve also had some experience with friends of mine who’ve tried getting the many American-built “parts guns” to work properly. You can damn near beat a functional AK-47 out of a former shovel, and there’s a guy out there who actually has. The folks trying to turn demilitarized parts kits into functional HK91 and HK93 clones do not have an easy time of it, and there are many who term it a black art. This set of facts does not argue at all for the supposed “ease of manufacture” with this sort of design.

Frankly, I think that if Mauser had had anything other than the 7.92X33 Kurz to work with back in 1945, they’d have gone in a different direction. That HK was able to parlay the roller-locking action for that cartridge into so many others over the years is more a testament to Teutonic obsessive-compulsiveness when it comes to attention to detail during precision manufacture than anything else. In other words, “ease of manufacture” in regards to roller-locking designs probably doesn’t apply outside of the narrow world of low-pressure, large-caliber cartridges, and German peacetime manufacture. There sure as hell haven’t been too many others who’ve managed to put similar designs into successful mass-production. And, I include the Swiss 510/StG-57 weapons in that characterization, as well. The Swiss are, after all, damn near honorary Germans when it comes to precision manufacture.

Hognose Post author

A handful of comments —

The Thais: sawadee, guys, but your line came with bundled engineers for a reason. Of course, that wasn’t quite the screwing Colt gave Chartered Industries of Singapore (they justified the purchase of production machinery on export sales, but their Colt license didn’t allow any).

CETME: they also couldn’t get the AMELI to work all that well. Completely different design but based on HK style roller locking

The American parts kits: Century is particularly responsible for these disasters. Barrels loose in the trunions != good headspace…. But there’s a big difference between building a gun in a factory and some bozo kitbashing milsurp parts together.


Which all goes precisely to my point-The HK roller-lock weapons are emphatically not easy to manufacture, nor does the design lend itself to anything other than a high-precision, extremely consistent industrial base for mass-production. The Germans managed it, but I have my doubts about how suited it is for anyone else.

The HK weapons are great, within their limitations. Those limitations seem to include “not being produced or supervised by the gnomes from Oberndorf”. And, given the paucity of recent roller-locked designs from HK, I’d also add a time frame in there, as well. I don’t know how true it is, but one reason I’ve heard for the transition from roller-locked designs to the current love affair with the AR-18 and AR-15 derived designs has to do with the retirement of key personnel at Oberndorf, and the difficulty of training new ones.

I could be wrong, but I’d be willing to bet that if they’d have gotten the StG45 into serial manufacture, it would have foundered under actual late-WWII production conditions. You never know, though-The Germans managed some other things nobody considered realistic, either. A lot of even late-war aero engines were machined and produced to tolerances we supposedly have trouble copying to this day. A chat with someone who is helping try to restore an ME-109 from late in the war was very educational for me, at least as to just what the Germans managed to do while we were bombing the living daylights out of the Ruhr.

Hognose Post author

109 was very unusual, and it was designed from the very beginning in 1935 to be easy to produce. The fuselage was designed in sections that could’ve been produced under sort of cottage industry conditions.

The bombing raids notwithstanding, German war production actually increased throughout the war and only finally begin to decrease in 1945, when Russian American and British troops began occupying the factories. That’s German ingenuity, that’s Speer’s brilliance, but that’s also pretty normal reactions to adversity by your standard issue human being.

The production engineering of the Me 263 was even more interesting, because it was put into production late in the war, and production machinery, personnel, and strategic materials like aluminum alloys, were all constrained. A lot of the 262 structure is sheet steel like a car body, welded. A project in the 90s reverse engineered the jet, and it was a bear to figure out.

I don’t believe any nation has an inside track on engineering. It tends to be at about the same level in every country at about the same time even when they are not communicating.

The state-of-the-art forces a good degree of convergent evolution.

Also, don’t underestimate the difficulty of manufacturing the A.R. 15. At least, manufacturing it to a quality standard. Even Colts had extreme problems in the early years maintaining items within specification, and they gave up keeping some things, like the magazine well, in specification whatsoever.