Friday, January 13, 2017

Against Mass Production in Post-Modernity: A Tractable Problem

There was a thing on Twitter, the other day to the effect that the Allies won World War II because Britain provided the time; the Soviet Union, the blood; and America, the treasure. I don't disagree, but I've been looking at various things of late. One is the shambles of the postwar British economy; the other, the M4 Sherman tank. The latter has treated me to a strong dose of the idea of American mass production. And while the M4 is a piss poor standard bearer for this argument given the collapse of production rates in 1944 after the completion of the capacity build out [pdf], the other corollary to the idea that mass production is that the Allies defeated better German tanks with human wave tactics, and, well, that's acceptable. And yet. It might be supposed that the Allies had an opportunity to end WWII in the fall of 1944 instead of the spring of 1945 by closing down the Normandy campaign with an even more complete victory than they, in fact, achieved. The larger consequences might or might not follow, but substitute a sufficiently good tank for the Shermans actually used in operations TOTALISE and TRACTABLE , and the larger operational victory was certainly in reach. Even in the final days of chaotic fighting around the Polish blocking positions outside of Falaise, a better tank might have made a difference. I've suggested before that Britain's last wartime cruiser tank, the Comet, would have made the decisive difference. But it seems at leasst worth arguing that the older Cromwell which would have replaced the Shermans of August in the absence of the 1943 production agreement would have sufficed.  "Good enough" is NOT good enough. 

This is a side profile view of the Sherman's precursor, the M3 Lee/Grant. It's intended to make you scratch your head at the weird machine and draw the inference that the US Army's Ordnance Service didn't know what it was doing in 1940.

As for the larger consequences, the official historians seem to think that ending the war a few months earlier would have had a salutary effect on the British economy. It would certainly have saved a great many lives. Is there a lesson about mass production here? Sure. Why not? It'll do for a subject this week, but I want to motivate this post a little more, so I'll briefly talk about counterfactuals, and the last stand of the BCRs, often written about in that way that demonstrates that historians do counterfactuals even when they think they don't.

Why would historians do counterfactuals? Because we do them all the time. At the end of OPERATION TOTALISE, a battlegroup consisting of an armoured regiment and mechanised infantry battalion, under the command of the BCRs' Lt. Col. Donald Worthington ended up stranded, deep in the German position, where it was destroyed during the daylight hours of 8 August 1944. The official history says that Worthington made an indefensible error of navigation on the difficult night march towards his actual objective, ending up six km away. He compounded this initial mistake by failing to acknowledge it, depriving his command of reinforcement and artillery support. This versin of events ought to motivate Vancouerites quite strongly, since Colonel Worthington was a scion of Vancouver elites (who lived two blocks from me), and the British Columbia Regiment is one of the city's two militia regiments. That disastrous last stand cost this city, and ought to have been a matter for the ballot box. See? There is a reason for counterfactuals! They tell us not to vote for the NPA.  (Not actually a disendorsement.)

There is, however, another version, in which Worthington was a gallant beau sabreur, who made an understandable mistake that ended with his command being in an excellent, if unforeseen position, from which, better supported, he could have led 2nd Canadian Corps to the closing of the Falaise Gap. Exactly why this support was not forthcoming is not clear, but David Bechtold very strongly implies that circumstantial evidence points up the chain of command.  then we need to look to higher headquarters, and have at the sacred cows of the Canadian army.

Whatever be the case, there is an interesting technical question here. I know, I know. No-one wants to hear some git prattling on about how, if the RAF had developed the Windsor, Whirlwind and Henley instead of the Halifax, Typhoon and Battle, Britain could have won World War II more! But when we can frame it as "Cromwells instead of Shermans SAVE THE WORLD," then the argument is motivated. (Becaue it's a critique of narrative tropes of mass production, you see.) If Worthington had better tanks, then TOTALISE might have accomplished what TRACTABLE set out to do. (Or, on the other hand, if all of 1st Polish had Cromwells, and not just the armoured reconnaissance regiment.) I'm narrowing donw the claim. Just the Cromwell, so close, in so many ways, to the M4.

All that in aid of saying that I like looking at pictures of tanks. I hope you do, too! This first one continues the established trope of talking about the Sherman by talking about every other American wartime tank.It's an absent presence! This tank that isn't the M4, is the Haunted Tank.  (M3 Stuart, for the inexcusably ill-read.)

CC BY-SA 3.0,

Does it look a little . . . strange to you? Hold that thought, because here's another American light tank of the same era, the M22 Locust, to show that American designers were perfectly capable of getting the light tank concept right.

Source: Warthnder Wikipedia.

Oh, sure, the Locust was a very specialised design. (It's supposed to be air portable, although it's not entirely clear how), but I would defend the claim that it is a reasonable comparison with the M3. 

On the other hand, the Locust was a late design. So, learning curves and all of that. Ann excellent point of comparison with the M3 would ber a first-attempt design by a team of engineers who had heard of tanks, but weren't exactly clear on what they were supposed to look like, do, or, really, anything, and, also, maybe got their engineering degrees as prizes in a package of miso soup. There are numerous candidates, but I went with this guy:
By Mark Pellegrini - Own work, CC BY-SA 2.5,

The Type 95 Ha-Go. After all, the Japanese were the nation who greased their machine gun bullets, thought that self-sealing gas tanks were for girls, and built any number of warships that kept trying to tip over because they the designers kept piling guns on top like they were working on Munchkin: Washington Treaty.Perhaps the comparison will help us understand that the M3 looks gawky for perfectly understandable reasons. Kind of like this table is probably going to look. 

Vickers Light Tank VI
Fiat M11/ 39
Type 95 Ha-Go
Light Tank VII Tetrarch
M-22 Locust
M3 Stuart
Alvis FV101
5.2 long tons
11.17 metric t.
7.6 metric tons
15.19 metric tons
Width, Length


Meadow 6-cyl gas
Fiat SPA
V-8 diesel
6cyl air-cooled inline diesel
2x6-cyl gas inline
Meadow inline 12 gas
6cyl opposed gas
Postwar gas or diesel
HP; HP-weight ratio
88hp; 16.9
105hp; 9.5
120hp; 16
Boy stuff: gun, max armour
15mm machine gun; 14mm
/46; 12mm
/85; 20mm
/52; 14mm
37/56.6; 63.5mm
Short 76mm; 12.7mm

All designs are compromises, and "light tanks" start out with the additional constraint of being, well, lighter than other tanks. That's why no Wikipedia entry for a "light tank" omits some kind of comment to the effect that the gun, or armour, or both, is inadequate compared with a regular tank. The reason for that is pretty obvious. The specification called for a light tank!

Leaving other reasons for manufacturing a light tank aside, the basic reason is that somethign is needed for the  reconnaissance/screening mission. "Lightness" in this context means the ability to evade. Depending on mission, this might mean sneaking past the enemy screen in order to carry out daring escapades. More likely, it means getting the hell away when you're spotted. Two specifications are important to scampering away without being seen. The first is the horsepower/ton ratio, about which I mean to remain a bit skeptical because of gearbox, braking and actual-horsepower-at-the-sprocket issues; the second, which I've telegraphed the holy fuck out of, is being inconspicuous. (Enjoy this highly relevant Monty Python short with the management's compliments.)

The "telegraphing" part, above, has taken the form of giving the tank's height a separate table row. You'll notice that the Russians, who know from tanks, managed to get the T-60 down to 1.75m; that the British tanks form a downward slope of height from 2.26 to 2.1m, and even those goofy Japanese deisgners have pulled off 2.13m.

The M3, on the other hand, is 2.56m tall. That's 7cm taller than the Cromwell, which was issued to British armoured divisions (and the Poles, but not Canadian 4th Armoured) as a reconnaissance tank, in lieu of more M3s or M4s. It is 12cm taller than the T-34. Although, to be charitable, it is shorter than the Panther (2.99m, although the third digit makes this grocer suspicious)!**

BCR veterans visited the hill of Worthington's last stand in the summer of 1945 and took these pictures, which the regimental museum (I think) subsequently provided to David Bechtold. In these frontals, the M4 has the verging-on-absurdity formal grace that some portly people take on. Look at that thrust-out belly, so unfortunately accentuated by the length of track, welded on in hopes of giving some small, additional margin of armoured protection. It turns out that it didn't really matter very much, as 63% of hits from in front were to the turret, and, in any case, virtually none failed to penetrate the M4's protection.

A slightly fuller summary of No. 2 Operational Research Group's examination of 45 Sherman tank hulls, shows that 40 had been penetrated by German 75mm or 88mm armoured-piercing shells, of which 33 had subsequently caught fire or brewed up. Of 65 hits by AP rounds, 63 penetrated the tank completely. (Summarised by Napier, but the full report is available in pdf here thanks to the work and generosity of Terry Copps, the doyen of Canadian military history.) No 88mm strike failed to penetrate. Since a great deal of emphasis tends to be placed on the sloped glacis of the M4's front hull, I will also note that fully 48% of penetrating strikes were more than 5 degrees off the normal angle of incidence, and that a total of 40% of strikes (from all directions) were against the turret. It's nice to have well-sloped armour, but an equivalent protection of unsloped armour is better.

That said, the Group's research shows that the sloped plate of the Pzkpfw V Panther's glacis was much more effective than vertical turret armour. Howevr, it also shows that hits against the forward glacis were far less common than against the turret and flanks, indicating just how rare it is for attacking tanks to get a look in at the forward hull of tanks in a defensive position. Worthington Force was, rarely enough for an Allied armoured unit in Normandy, fighting a defensive battle.

Notice, by the way, that apart from the M4's thin armour, it was frequently suggested at the time that the M4's armour also had poor resisting power for its thickness. This issue has not been pursued postwar, although the operational group notes that reissued, reconditioned tanks might well have poor quality armour dute to the stress of previous penetration.

So the M4 doesn't just look strange and vulnerable in this amazingly rare full frontal  view (I couldn't find an equivalent for the Cromwell, so here's something close:

Source: That's pretty unsloped armour, admittedly.

 It is vulnerable. It's a big target! In a fire fight, you want a small target! That's all I'm saying, although a bolder claim would be about who spots who, first. The takeaway here, not to bury the thesis or anything, is that if Colonel Worthington had had 2.49m tall Cromwells with 100mm armour on the turret front instead of 2.74m tall M4s with 76mm armour, he might well have done a bit better in the fight itself. It is also likely that he would have got more of his original command through the approach march. And, in a perfect world, the accompanying forward observer for the attached 5.5" artillery regiment wouldn't have suffered a mechanical casualty in the approach march. Or, at least, would have been able to cadge a replacement with an adequate radio, allowing him to call in artillery support. (Actually, taking the whole battle into account, better radios probably would have done more than better heavy metal, but what are you going to do?)

Why? Well, you'll be glad to know that I'm not tired yet of talking about the M4 by talking about every other American tank ever built. I started off with a profile image of the M3 Grant/Lee, the first American attempt at a medium tank. Here's another look.

I also now need to take back, on record, every mean thing I ever said about the Japanese engineers behind the Ha-Go. This is a tank? It's 3.12 freaking meters tall! The British Purchasing Commission made the Ordnance take the commander's cupola off to bring the height down a bit, but inconspicuous, it ain't.

This is the M60A1:

Design work for the M60 began in 1957, well after the Centurion had appeared in its final form, and, in fact, after work on the Chieftain and Leopard I had begun. You will notice that the commander's cupola is back. It will be removed, finally, in the A2 variant, but, in the mean time, the M60 is 3.213 meters tall. The T-64, by way of contrast, is 2.172 meters. More than a meter shorter. With this in mind, it is very hard not to conclude that the return of the cupola to the M-60 isn't just trolling us. Someone at the US Ordnance simply does not think that tank height matters. He's wrong, and so wrong that you have to wonder what the heck is going on.

Back to the M-3.You will notice that I've given a link in lieu of details of the engine. This is that engine.

By FlugKerl2 - Own work, CC BY-SA 3.0,
It is the Continental Aviation and Engineering Company's R-670, a perfectly workmanlike 11L, 7 cylinder, air-cooled aircraft engine, in what is known, for obvious reasons, a radial configuration. Radials are a common aircraft engine type, especially for training and transport applications, but used in high performance roles as well. The key advantage of this configuration is that the cylinders are as far separated as possible, in order to achieve the best possible cooling air flow.

Now here's the M4's engine.

This is the Wright Whirlwind R-975, a 9 cylinder radial with a 16L displacement, giving 420hp, although most of the ones that went into tanks were built under license by Continental. The circular rim is the shroud that keeps in the cooling air, blown across the cylinders by a fan running off the crankshaft, which is another of the many reasons that a tank's reported hp/ton ratio isn't to be taken entirely seriously.

The decision to give Continental a tank engine production contract was not a bad one. The company started out in aviation, but the aero-engine business was relatively small, so in the 1920s it branched out into making car motors for several companies that failed in the 1930s. The company was left with a large and relatively modern, shuttered factory near Detroit. Since Continental had no new aero-engine designs even close to being ready for production, getting it into the tank engine business just made good sense. The decision that I am questioning here is the one to order a radial engine. The Fisher Arsenal, built by Chrysler specifically to make tanks for the US Army, was not allowed to source Wright or Continental motors, and improvised a replacement by coupling five of its existing six cylinder inlines into what Wikipedia gently calls a "unique" 30 cylinder, 21L configuration. It required a longer hull, but since that doesn't impact frontal area and thus the weight of the main protective armour volume, we'll let it pass, especially since Chrysler engineers managed to get the thing working in only a year, and it proved very reliable in service. (Although I cannot believe that serviceability statistics include a charge for tune-ups, considering the amount of work it must have taken to change and regap 30 spark plugs.) Note that not only Chrysler, but also Ford was able to produce a conventional engine for the Sherman; and there were also diesel-engined Shermans, running the General Motors 6046, a doubled Series 71. 

The official history of the Ordnance Service says that, prior to mobilisation, Ordnance tank designers intended to use the Guiberson diesel or Wright Whirlwind as standard tank engines. [pdf again: 244). The Guiberson, unbelievably, was a 9-cylinder radial, air-cooled diesel aero-engine, and a less-promising technological concept it is hard to imagine. Mr. Guiberson, whoever he was, contracted the manufacture of his brainchild to Buda Engines, an also-ran internal combustion builder of Illinois which was mainly building other firms' designs under license when it was finally taken over by Allis-Chalmers. Along the way, though, it made a fair number of --Jesus Fucking Christ-- air cooled radial diesels to put in tanks.

That's why the M4 is so big in the front. It was designed around an engine concept whose elevator pitch is: "This configuration takes up the most room of any possible, practicable engine configuration." Note that this decision wasn't taken in haste, and reflects a coherent philosophy. All three of the American tank designs developed during the war were originally designed to take some kind or other of radial engine. A comment, deep into the chapter, to the effect that the committe had not yet decided on whether to favour air-cooled or liquid-cooled engines probablly gives the game away, and, curiously enough, the M60 had an air-cooled engine, long after the "controversy," such as it is, ought to have been settled. The American Ordnance Service was deeply, deeply commited to air-cooled engines for AFVs, and, apparently, had no problem with making the armoured box bigger, thus heavier, to accommodate it. I'm not crazy in suggesting that this wa a mistake, that liquid cooling is probably the way you want to go, just on the basis that that's what everyone else does, right?

The official history is a bit oblique in giving credit for this admirably consistent, if incomprehensible choice is a person identified as "Brigadier General Christmas,"or, in the index, as Brigadier General John K. Christmas. General Christmas has done an excellent job of evading fame. The name is uncommon, so all I can tell  you about the man is that he might have family. There is a retired Marine Corps general of that name, born in 1940. And, of course, a contender for the much-coveted, highly competitive title of "Worst Aircraft Designer Ever," Dr. William Christmas.

The Christmas Bullet.

Dr. Christmas might have had some pull with the Ordnance Service, considering that he got them to loan him a Liberty 6, although Out of the Box credits a senator's intervention as being decisive. There's really only the coincidence of the name to connect them, but it would be appropriate, somehow, in the end, if the driving force behind the M4 were to prove to be connected to this quintessential American patent troll/grifter.  Dr. Christmas abandoned the field of medicine in favour of aeronautics early in the century. Considering that he not only ignored all advice about leaving the Christmas Bullet's wings unstrutted, so that they could "flap like a bird's wings," but put a second pilot in the air in an another prototype after the first had been killed in the initial test flight due to a bad case of wings coming off, we can imagine what kind of a doctor he was. "Seldom right, rarely in doubt," etc.

Someone at the Ordnance Service, it very much looks, was of the same personality type.
Just sayin', is all.

At its worst, "mass production" is a grift. The basic claim of mass production is that, by standardising a design, you can streamline production processes to produce more of a given item. In the case of the M4, it is certainly true that a great many were manufactured. During the production period, the United States may have produced three times as many medium tanks as the British, and even taking into account the production rundown in Britain from late 1943, this is impressive, though nowhere near so impressive as the 83,000 T-34/76s and T-34/85s produced.

It is, however, the basic claim that goes wrong. From first to last, the Fisher plant made its own Sherman. The engine was unique, the hull was unique, and one can add in different armaments, which included the basic 75mm, failed 76mm high velocity, 90mm and 105mm guns; as well as different combinations of cast, rivetted and welded hull components. In total, twelve different manufacturers produced 7 different Army-recognised variants, while the "Sherman Vc" identified below is a Royal Ordnance designator for an after-product variant. Wikipedia counts five different engines, and the "E" subvariants produced in late 1944 were in some ways more distinct from the other "A variants" than the "As" were from each other.

Given this range of variants, one might expect a version of the common manufacturer's complaint, that production quotas could not be met due to the frequency of change orders and design variations. In fact, however, the crisis of production that left the re-equipped 4th Canadian and 1st Polish armoured divisions at less than 85% strength for TRACTABLE compared with TOTALISE is due to the programmed American production facilities running at half capacity through 1944-45. The reason for this, as for all of the late war munition shortfalls, is clear enough. After showing a remarkable ability to increase employment levels in the early years of mobilisation, the American workforce had run out of steam, and there were simply not enough workers to staff the factories.

The myth of American mass production is, at some level, the same as the myth of the Red Army steamroller. Somehow, the usual rules of demographics are suspended, and America is capable of producing as much as it wants, just as the Soviet Union is capable of fielding an army that ought to be 30 million strong (600 divisions x c. 50,000 head divisional slice). Pressed on the numbers, we resort to unexamined myths. The Soviets don't need mechanics and labourers and replacements because they're quasi-Asiatics, and, you know, Asians. American productive capacity is not limited by population because, you know, Americans. ("Mass production.")

But there is another aspect to this myth. Mass production works because "Mr. Ford would produce a car in any colour the customer wanted, as long as it was black." Customer choice is a silly luxury in the face of the efficiencies of mass production!

2nd British to 5/08
Recovered nett
1st Canadian to 26/08
2nd British 6-26/08
Total 21 AG
US 12th AG
Gross total
Sherman M4
Sherman Vc


Churchill (Churchill 95)

150 (+11)
Table lifted from Stephen Napier, Armoured Campaign in Normandy, June to August 1944 (n.p.: History Press, 2015): page number It's a Kindle Secret! (But it's in the summary chapter.)

The Allies lost 2000 M4s in Normandy. It would be very, very hard to make a case for one tank over another in Normandy, but there were 5 armoured divisions, 5 independent armoured brigades, and 3 army tank brigades under 21st Army Group, exclusive of the 79th Armoured Division, the main home of Churchill variants. That's 28 Sherman regiments, 7 Cromwell, 9 Churchill. On the evidence to hand, and subject to all sorts of qualifications about period in action, it sure looks as though gigantic, lumbering, thin-skinned tanks were not "good enough."

Could it have been avoided? Sure! Adopt British designs. Oh, sure, put an American engine in, instead of the Meteor. The Ford one sounds like it would have been good enough, and I am sure that American industry could have matched the Bedford in the Churchill. Yes, this would have meant changing the tank design four times in the course of 1940--45. Oh My God! More design, less production! It's the end of the world! But 2000 tanks. Come on. Since when is it an American philosophy of war to defeat the enemy with human wave tactics?

So, yes, the M4 was good enough to win the war. But here, more than perhaps anywhere else in the conversation about technological counterfactuals, it is reasonable to suggest that a more, dare I say it, British-like production approach would have won the war more.

Also, I can't help but notice that while I can't get a Palm, Nokia or Blackberry smartphone any more, I can get an iPhone in black.I can also get a cheap Android, admittedly. The tech writers, who had a chance to look at the late generation Palms and Blackberrys, say that they actually had a better operating system than Android. But, you know, who cares, something about cheapness, or apps? Also, Steve Jobs was a genius.

*Surcouf, Steam Gunboats, USS Alaska, pocket battleships, Kitakami . . . Hey, that could work!
**No, we don't use ".99" pricing to make things seem cheaper. It's for counting off quantity errors at the till.

No comments:

Post a Comment