Landships II

Members Login
    Remember Me  
Post Info TOPIC: Canon de 75 mle 1897/ US Model 1897 recoil system function and manufacturing.


Status: Offline
Posts: 31
Canon de 75 mle 1897/ US Model 1897 recoil system function and manufacturing.

Since I found a 1920 article on the manufacturing process of the French 75's recoil system by Singer, I decided to clean it up a bit (removing typos and converting units to metric) and post it, which will be on this thread.  But in order to understand the significance of this, it is important to know how the recoil system works and why it was so hard to make.

There are a few pictures of the French 75's recoil system floating around the internet:

However, these do not accurately show how the system operates, because they are based on pre-1918 public diagrams with all the details censored out (the diagram in the second link is an actual pre-WWI image).  This can be seen on page 273 of Henry Arthur Bethell's Modern Guns and Gunnery, published in 1910, which shows the same general layout and states that "The exact details of the recoil gear are kept secret, but it is understood that the arrangement is in principle as shown in the above diagram."

The full mechanism is seen in more recent technical documents, such as the US Army's Recoil Systems Metric, from 1988 (It's a download, so don't click unless downloading it is intended).  Pages 3-9 and 3-10 show and explain this diagram:


Though it does not state that this was used in the French 75, it is labeled the "Puteaux mechanism" as it was invented there for the French 75.  The book "Elements of Ordnance" by Thomas J. Hayes, from 1938, shows the same mechanism on page 259 and clearly states that it is used on the French 75.  

Using these 2 sources, the functioning of the recoil system can be explained, using a series of simplified diagrams:


Puteaux Mechanism diagram 1

In the top diagram, the recoil system is at rest.  Dark blue is the hydraulic fluid, and light blue is air.  The orange piston is attached to the gun barrel.  In the bottom diagram, the first part of recoil is underway.  The gun recoil (black arrow) pulls the piston (orange) backwards.  This pushes the oil down into the lower cylinder where it then pushes the floating piston/control rod (green) forwards.  The oil flow is represented by the red arrows.  It freely flows into the lower cylinder, and then flows between the orifice (purple) and the control rod (green) as it pushes the floating piston forwards.  A little bit of oil flows right around the control rod (the 2 very small arrows) instead of through the 2 big ports on either side, but it's not much by comparison.

 Puteaux Mechanism diagram 2

In the top diagram, the recoil stroke has progressed further.  The piston (orange) is still being pulled backwards, and the oil still is flowing into the lower cylinder (red arrows), but it can't move as fast anymore.  As the oil pushes the floating piston/control rod (green) further forwards, the fat end of the control rod fills up more and more of the orifice (purple).  This means that the oil can't flow into it as fast, and it has to slow down- which means the piston (and gun) has to slow down.  This absorbs the recoil energy.  In the bottom diagram, the gun is at full recoil.  The piston (orange) has been pulled all the way back, and the oil has pushed the floating piston/control rod (green) so far forwards that the control rod completely blocks the orifice (purple).  This means that the oil can't flow forwards at all, and the piston can't move backwards at all.  At this point, the gun has been brought to a complete and gentle stop (because the control rod slowly and smoothly blocks the area).



Puteaux Mechanism diagram 3

In the top diagram, the recoil system is returning the gun back to position (counter-recoil).  The compressed gas pushes the floating piston/control rod (green) backwards (black arrows), which in turn pushes the floating oil back into the upper cylinder (red arrows), and which pushes the piston (orange) and gun back forwards.  If this happened the same way as the recoil in reverse, the oil would flow faster and faster as the control rod blocked less and less of the orifice (purple).  This would cause the gun to slam into the front of the recoil system at high speed, and jump forwards.  So to stop this, a set of one-way valves (dark blue) block the main passages, and force the oil only to flow through the small space around the control rod (the very small red arrows).  This forces the oil to flow slowly, and push the rod (orange) and the gun forwards slowly so it won't jump forwards.

The bottom diagram shows the actual nature of the floating piston in the Puteaux mechanism.  The control rod is attached to a diaphragm, which is attached to the actual floating piston (green) by a spring (light green).  This is so that the control rod is always pressed against the orifice in the right spot when the recoil starts.  If there is an extra amount of oil in the system, the floating piston gets pushed further forward and the space between the piston and the diaphragm gets bigger, where the extra oil is stored.  This way the floating piston won't pull the control rod too far forward when there's too much oil in the system.  When the mechanism leaks oil, the compressed gas pushes the piston further back, that extra space just gets smaller to account for the lost oil, and the recoil system functions normally.  This is why the oil in that space is called the oil reserve- it's to compensate for oil leaking out of the mechanism.


What's notable about this system is that it was exceptionally hard to make in WW1, even though plenty of hydrospring recoil systems were mass-produced just fine at the same time.  It required extreme precision, as several sources show:

One paper, "Evolution of the American modern light field gun" from 1978, states on pages 49-50: Upon close examination, the secret of the Puteaux recuperator was revealed. Each one was hand made to an indescribably close tolerance with precision nearing perfection. These extremely close-fitting parts and highly machined surfaces could not be adapted to assembly line production needed to quickly produce guns. This is the reason the British elected not to use hydropneumatic recuperators until 1918 when they developed proper manufacturing techniques. The Germans never attempted to produce the recuperator during the war after examination of captured French guns.  The exacting construction of the recuperator also posed a significant problem to the Ordnance Department after the decision was made to produce the French gun.

Another comment on a forum quotes from Ian V. Hogg's book "The guns 1917-1918," stating that the secret to their success was "Fine tolerances, and an exceptionally close-fitting piston-head sealed with german-silver rings. Nothing more."

Since hydrospring mechanisms are only different in having a spring rather than compressed gas, logic shows why this mechanisms requires such high precision.  A hydropneumatic recoil system has 3 types of seals that might require high tolerances: those separating oil from oil (like the one-way valves), those separating oil from outside air (like the gun piston seals), and those separating oil from compressed gas (like the floating piston).  A hydrospring system only has the first 2 types of seals.  If we look at the consequences of each seal leaking, the following happens:

  • If a seal separating oil from oil leaks, a negligible amount of oil will move around the mechanism- this is negligible compared to the normal flow of oil through the mechanism anyway.
  • If a seal separating oil from outside air leaks, the oil is under much higher pressure than the air, so oil will leak out, but air won't leak in.  There are oil reserves in the mechanism to compensate for the oil that leaks out, so this isn't a problem.
  • If a seal separating oil from compressed gas leaks, the oil and gas will slowly mix.  This means that not only will the compressed gas become less compressible because of the oil (and be a less effective spring), but the oil might become more compressible because of the gas in it (and compress when pushed instead of flowing to slow the gun recoil).  This is a serious problem.

So what makes the Puteaux mechanism so hard to make is likely that floating piston separating oil and compressed gas.  It can't be allowed to leak under any circumstances, which means it would need those high tolerances to form a perfect seal.  (This would also explain why Schneider guns were easily mass-produced in WWI despite also being hydropneumatic- their recoil mechanisms didn't have floating pistons separating the oil from the compressed gas).

Some other articles and sources bear this out, stating the floating piston was the main secret of the Canon de 75's recoil mechanism.  This article states: The floating piston was of particular interest to those wanting to copy the gun’s design because of the way it was sealed to prevent the fluid and gas from mixing. This was such an important detail that French artillery officers were forbidden to have any knowledge of it—in fact, they were not allowed to see the piston itself when it was disassembled from the gun. Various regulations were put into place to assure the secrecy of the 75’s internal mechanism.

More importantly, the US Army M1897 manual from 1942 has no repairs for the floating piston.  Most problems apparently have some fixes that can be attempted in the field, but only one problem involves the floating piston, on page 54: If the presence of air in the recoil oil is due to the escape of nitrogen past the floating piston, the recoil mechanism must be returned to an arsenal for repair.  Apparently the floating piston was so sensitive that it could only be properly maintained in an arsenal.

So, with that background, the importance of the manufacturing process for this gun should be clear.  I will end this post with one more quote, from another article about the Canon de 75 (by Robert L. O'Connell):  A break finally came when, under some­ what murky circumstances, the artillery officer training program received four worn-out 75s as training tools. The guns soon found their way to the Ordnance Department, where they were torn apart. After disassembling the guns, the army’s chief of artillery was telephoned and told that the secret of the 75’s recuperator was nothing more than incredibly close tolerances–no tolerances, actually. Each and every example had been handmade by French craftsmen, working with the precision of jewelers.  

Characteristically, the Americans chose mass production even after the French warned them it was impossible. It took time (in fact, the war was over), but the Singer Manufacturing Company, with ex­hausting attention to detail, finally man­aged to fabricate acceptable recuperators based on machine-made interchangeable parts. According to artillery historian Frank Comparato, the project “was one of the most complicated engineering tasks ever attempted…one of the marvels of modern industry.” Singer’s employees had every reason to celebrate. It is not known whether the French, when they heard of the feat, shared their joy.


It is this manufacturing process Singer developed that will be described in the next posts.


-- Edited by AN5843 on Saturday 7th of December 2019 12:12:02 PM

-- Edited by AN5843 on Saturday 7th of December 2019 12:12:42 PM

-- Edited by AN5843 on Saturday 7th of December 2019 12:13:32 PM



Status: Offline
Posts: 2281


Secrets don't last forever. After WW1 the French Army published an "Atlas of Lithographs of the Canon de 75 Mle 1897".

The details of the floating piston and recoil system are shown in great detail.






Status: Offline
Posts: 31

CharlieC, that is indeed an excellent source, and shows the mechanism well.  As mentioned below, that's actually one of the sources I used to confirm whether the article was actually about the Canon de 75's recoil system.

Below is the 1920 article about the manufacturing of the Canon de 75's recoil system by Singer.  It is from Volume 52 of American Machinist, and titled "Unusual Methods of Securing Extreme Accuracy," by A. L. De Leeuw.  It is a 4-part article which can be found at these sources:

Google Books source 1

Google Books Source 2

Internet Archive Source

Part 1 is on page 595, Part 2 is on page 937, Part 3 is on page 1049, and Part 4 is on page 1094 on any of those sources.  You can use those sources to read the whole article right away if you want.  

I have copied the figures from the Internet Archive source, as it has a better quality scan.  I corrected typos and errors, added metric values in parentheses in red text, and added blue letters and arrows over the original arrows and letters in figures (which made them easier to read).  The figures were moved to more suitable locations in the article (they are normally far from where they are mentioned in the writing), and I added some writing to the captions, in parentheses in red text, to make the figures easier to understand.

One interesting thing about the article (and why it was so hard to find) is that it never refers to the 75 mm gun- it only states that it is a recoil mechanism.  However, we can determine what gun the recoil system is for by 3 methods:

  1. The author, A.L. De Leeuw, is cited in several other sources (such as this one) as being in charge of Singer's efforts to manufacture the Canon de 75's recoil mechanism, and in some of his other writings he mentions first using certain milling operations (which can be seen in the article) on the 75 mm French gun.
  2. The article mentions the work was done at Singer, and the previously mentioned sources state that De Leeuw was the chief engineer there.  Upon checking contracts with the US Army during WW1 (page 60), it turns out that the only recoil system Singer made was the Canon de 75's.
  3. Just looking at the figures in the article and comparing them with actual diagrams/pictures of the Canon de 75's recoil system (mainly that on page X of CharlieC's link above) shows it to be identical.
  4. The tolerances mentioned in the article (0.0008 inches- 0.00025 inches achieved in practice) fit the reputation of the Canon de 75's recoil system for needing high precision.

So with those methods, it is clear that this article describes the Canon de 75's recoil mechanism.


And with that, I can start posting the article:

Unusual Methods of Securing Extreme Accuracy—I

Consulting Engineer 

Accurate drilling and boring of long holes is one of the greatest of shop problems. When two holes must be bored parallel, the problem increases in difficulty. This article tells how the preliminary machine work necessary to the locating and boring of such holes was per formed. The illustrations show the problems clearly. 



(This image has no designation; I just refer to it as Fig. 0.  It shows the final shape overlaid on a rough forging, like Fig. 3 below.)

The problem of making an accurate recoil mechanism for gun carriages is a particularly difficult one, both on account of the extreme accuracy required in the component parts as well as owing to the very large amount of metal to be removed from the main, or cradle, forging. This main forging was supposed to weigh 950 lb. (430.913 kg) in the rough and 215 lb. (97.5224 kg) after machining; but, due to the lack of satisfactory heavy forging machinery, it weighed from 1,300 to 1,350 lb. (589.6701 to 612.3497 kg) and it did not seem possible to reduce this weight and still make a forging which could be completely turned up. Fig. 1 and some of the other views show the irregularities in the forging. The waves or irregularities were sometimes as much as 1 1/4 in. (31.75 mm) in depth.

The unfinished appearance of the trunnion lugs, and the curved outlines of the forgings can readily be seen. Fig. 2 shows the large amount of metal allowed, but it sometimes happens that even with this it was barely possible to true up the trunnion. Fig.3 shows the cross-section of the forging to which has been fastened a steel templet showing the cross-section of the finished part; the two holes shown are the result of the first operation.


 Fig. 1.  The irregularities of the forgings



 Fig. 2.  Surplus metal on trunnion



 Fig. 3.  Metal removed at various operations

Fear was expressed that, due to the removal of such a large amount of metal and to the delicate shape of the product, there would be great danger of twisting and warping of the forging after machining. It was found, however, that with the sequence of operations adopted, there was no trouble from this source. As a matter of precaution it was decided to leave a small amount of metal on the inside of the slide part to be removed after finish-boring; and, though this was regularly done, it was more than doubtful whether this operation was really necessary.


As the forgings came so rough and of such uncertain dimensions, it was necessary to lay out every forging very carefully before machining. This at first required the services of a skilled man and consumed a great deal of time. As the plant was originally laid out for a capacity of 25 finished recoil mechanisms, and required 30 of these finished forgings per day to allow for waste, a machine was designed to do this laying out quickly, accurately and with unskilled help.

This machine is shown complete in Fig. 4 with details in Figs. 5 and 6. It consists of a bed, and means of shifting the forging as desired and for finally clamping it in the proper position. To obtain this position, brackets were used which could slide in grooves in the bed, these brackets carrying feeler rods which could be placed by hand, in vertical or horizontal directions, against the forging. If these rods touched the forging in any position of the brackets on the bed, then the forging would have sufficient metal for finishing at all points. If one of the feeler rods, when placed over or along the forging, failed to touch it, then the forging was shifted horizontally or vertically, or turned around its axis, until the rod struck.

At each end of the bed a drill head was mounted, the head containing two drill spindles, each having an independent belt drive. The feed was by hand. The holes drilled in this position corresponded with the center of the holes which finally would have to be bored through the forging and were small enough so that it was not necessary to drill them with extreme accuracy. These holes were the starting point of all the subsequent operations. Fig. 5 shows the construction of the feeler rods and brackets. The method of locating the center for the trunnion parts by means of center punch is shown also. Fig. 6 shows the drill heads and jig brackets.



 Fig. 4.  Machine for laying out forgings.




Fig. 5.  How the machine is used.




Fig. 6.  Drilling the two pilot holes.  (Figs. 4, 5, and 6 show the same machine)



Status: Offline
Posts: 31

I'm probably going to upload Part 1 in 3 posts- with the above post being the first.  Figure 7 in particular will probably make up most of the second post, as it's a huge image.



Status: Offline
Posts: 31

On to the second piece of Part I:


The sequence of operations is shown in Fig. 7. The milling of the two guide strips at the bottom of the forging, operation 2, is shown in Fig. 8, where the piece is set up for operation. This operation was performed on a 24-in. (609.6 mm) milling machine with an inclined rail. The illustration clearly shows the jig and holding devices, and especially shows how the holes drilled in operation 1 were used for locating the forging for this operation. The cutting speed for this operation was 57 ft. per minute (17.3736 meters per minute), the feed 0.78 in. (19.812 mm) and the time for finishing a piece about 2 hr. Before the milling machines were completely installed, this process was done in a planing machine, requiring about 8 hr. per piece.

After this operation the lips thus milled and the surfaces thus obtained were used as gages or control points for further operations.





 Fig. 8. Milling the locating surfaces.

Fig. 9 shows operation 2A—rough-milling the sides and the stock around the end of one trunnion. This operation was not contemplated in the original lay out but the additional amount of metal of the forging made it necessary to remove some of the metal before the finishing of the sides. The method of holding the piece is clearly shown, including the hardened-steel plates against which the lips of the forging are located before milling. This operation was done on a No. 5 milling machine and required two settings, as the trunnion lug, which is practically in the center of the forging, would not allow of completing one side in a single setting. The milling machines were consequently arranged in pairs, one running right hand and the other left hand, so that one pair of milling machines could take care of a complete side. The depth of cuts varied widely with the forgings, being some times as much as 1 7/16 in (36.5125 mm). The feed was 1 1/4 in. per minute (31.75 mm per minute). The style of cutter used was the 8 1/4-in. (209.55 mm) high-power face-mill with an extra amount of projection of the blades beyond the body. Operation 2B is the same as the previous one but on the other side of the forging.



Fig. 9. Removing material from side. (The same is done on both sides)

Operation 3 is shown in Fig. 10 and consists of shaping the top of the forgings. This was done on a 24-in. (609.6 mm) horizontal machine with inclined rail and interlocking cutters. The cutters were kept as small as possible but were nevertheless 8 in. (203.3 mm) in diameter. This was caused by the fact that the arbor was 3 in. (76.2 mm) in diameter and further, that a deep section of metal had to be cut. In these cutters and in all gangs of cutters for heavy work, a cylindrical key, as shown in Fig. 11, was used. With this key no trouble was experienced in getting the cutters on or off the arbor.



Fig. 10. Milling the back of forgings.



Fig. 11. Key used in milling cutters

-- Edited by AN5843 on Wednesday 11th of December 2019 05:35:37 AM



Status: Offline
Posts: 31

The third and final post of Part I:


Fig. 12 shows operation 4 which was done on a double spindle 24-in. (609.6 mm) machine with vertical heads, using a cutter which was 5 in. (127 mm) in diameter and 4 in. (101.6 mm) high. The illustration will show how extremely heavy this cut was. It also shows the peculiar style of stub arbor used, as it could not be expected that any ordinary kind of arbor would stand up under this heavy duty.


 Fig. 12. A heavy milling cut.


By a misunderstanding of the instructions, the men in the shop divided this operation into two cuts, and invariably broke the cutter or arbor, or both, on the second cut. When, however, the entire amount of metal was removed in one cut there were no breakages. This was due to the fact that the metal to be removed lying immediately back of the body of the cutter prevented the cutter from pulling itself into the cut. It will be noticed that stops were provided on the cross-rail to keep the heads from moving under the cut. Two of these are shown at A and B. The illustration shows clearly how the pieces were clamped and how two pieces were done at one setting. It also shows the steel water guard around the table.

Fig. 13 shows operations 5 and 6 which are different only in the shape of the section of metal removed. Operation 5 was done with a relieved cutter held by means of a stub arbor and used as an end-mill. Operation 6 was done in a similar way except that the cutter was made in two parts and interlocking. The reason for this difference was that the surface made by operation 6 had to be corrected within narrow limits, whereas the surface made by operation 5 did not require such accuracy and the variation of thickness of the cutter due to sharpening would not cause trouble. It was found necessary to introduce an extra operation before operations 5 and 6. This operation is shown in Fig. 7 as 4A and consisted of beveling the end of the piece where the cutter is to enter. Without this precaution the cutter was apt to pull itself into the cut, though there was no danger of such a mishap after the cutter had once entered. 


 Fig. 13. Undercutting the sides.



Operation 7 consists of removing practically all of the metal from the inside of the forging. This was done on a 24-in. (609.6 mm) horizontal milling machine, using interlocking cutters. As it proved to be extremely difficult to get sufficient cutters for this operation, and as the first cutters obtained were not correctly made, it was decided to split this Operation, for the time being at least, into operations 6A and 7.

It will be noticed that operation 6A can be done with ordinary milling cutters, which of course, could be made in a relatively short time, and without the use of a large backing-off machine. This left much less metal to be removed by the special cutters, and made it possible to do the finishing operation, No. 7, at higher speed and feed and with less wear of cutters than if the entire amount had been removed in one cut. However, if cutters had been available, it would have been more economical to take one single cut with cutters as in Fig. 14.


 Fig. 14. Milling out the channel.


Operations 8 and 9 are the rough-finish-milling operations along the sides of the piece and around the trunnion, Fig. 15. These operations are similar to operation 2A except that less metal must be removed and consequently a smaller cutter can be used. This operation was done on a No. 5 machine and as the table travel was only 50 in. (1270 mm) and the length of the piece to be milled from 69 to 70 in. (1752.6 to 1778 mm) it was necessary to mount the piece in a sliding fixture. Fig.16 shows the cutting of the lip of the forging by means of a relieved end-mill. This lip is made of very delicate shape and to men not versed in the mysteries of ordnance design, it would seem that the gun would shoot equally far and straight with a square lip. However, as the design had to be followed, the special milling cutter was used on a No. 5 vertical milling machine and the piece was mounted on a sliding fixture which is quite clearly shown. 


Fig. 15. Finishing the sides



 Fig. 16. Undercutting the lip


Operation 11 requires the removal of a considerable amount of metal; only 1/32 in. (0.79375 mm) was left on the inside of the forging for finish-planing. This can be seen in Fig. 7. The central guide running lengthwise of the forging limited the size of the cutter, as it was further necessary to make some little allowance for the sharpening of the cutter and the consequent reduction in diameter. This, on account of the forging, limited the stem of the cutter to not more than 1 1/8 in. (28.575 mm). As it was evident that an arbor of 1 1/8 in. (28.575 mm) in diameter could not possibly resist such a heavy cut, the arbor was made 1 1/8 in. (28.575 mm) only at the point where it passed the lip of the forging; while immediately after this point was passed, the arbor widened out and followed closely the outline of the forging so that the portion which was 1 1/8 in. (28.575 mm) in diameter was only about 3/8 in. (9.525 mm) long. This proved so successful that no breakages occurred.

This cut was taken in a No. 5 vertical milling machine using a sliding fixture. Operations 12 and 13 are shown in a general way by Fig. 17. These operations consist in milling off the end of the forging and then milling a gap through the center part. This gap was 2 in. (50.8 mm) deep at one end of the forging and 4 in. (101.6 mm) at the other. After these operations, the forgings were 69 in. (1752.6 mm) long while the length between the gaps was 63 in. (1600.2 mm). This operation was done on a No. 5 machine with an end-mill. 


 Fig. 17. Slotting the end



In doing such extremely heavy work it was necessary to drive the cutter with a key at the end of the mill rather than with a regular key on a stub arbor. The regular key has a tendency to split the mill. In all these operations the fixtures were provided with setting pieces which were 0.010 in. (0.254 mm) below the desired surface, and feelers with a thickness of 0.010 in. (0.254 mm) were used for setting the cut. In cases where it was necessary or advisable to run the machine while setting the cutter, a feeler of copper was used. However, this was avoided as much as possible for safety's sake. In all milling operations large amounts of cutting compound were flowed over the cutter.



Status: Offline
Posts: 31

Here's the first post of Part II:

Unusual Methods of Securing Extreme Accuracy—II


Consulting Engineer 

In this installment the very important operations of drilling and boring the two main holes in the cradle forging are taken up in detail. The methods used, the types of tools and the working limits are all discussed.
(Part I appeared in our March 18 issue.) 

The machining operations which are the most important in the schedule of operations on the cradle forging are the drilling and boring of the two main holes. The entire success of the mechanism depends on the accuracy with which these holes are bored. These holes are 63 in. (1600.2 mm) long, one is 1 9/16 in. (39.6875 mm) in diameter, and the other about 2 1/2 in. (63.5 mm). The variation permissible in any one hole is 0.0008 in. (0.02032 mm), though the permissible variation between the holes of various pieces is as much as 0.002 in. (0.0508 mm). However, this allowance was of no assistance as the pieces were supposed to be interchangeable. There was nothing in the mechanism which required the two holes to be parallel, but, as very close limits were set on the walls surrounding both holes and as the metal between the two holes was rather thin and had to be submitted to a hydraulic pressure test of 5,700 lb. to the square inch (39300.12 kPa), it was thought best to make the holes parallel whether they were required to be so or not.

In addition, the holes were to have a very smooth and mirror-like finish and were to be entirely free from tool marks and even the smallest surface scratches. The result of the following operations was that the variation in individual holes did not exceed 0.00025 in. (0.00635 mm), that the roundness of the holes did not in any place show a measurable variation and that the surface was sufficiently good to withstand the very severe tests to which the pieces were subjected, of which there were more later on.

The boring operations took place in the following sequence: Drill small hole; drill large hole; first bore small hole; second bore small hole; first ream small hole; second ream small hole; first bore large hole; second bore large hole; first ream large hole; second ream large hole.

It was the original intention to drill the holes on special 30-in. (762 mm) boring lathes. These lathes had a 10-hp. (7.457 kW) motor mounted on the carriage for turning the drill and a revolving steadyrest for the work or fixture to revolve in. Delayed delivery of these machines made it necessary to rig up other lathes, originally designed for boring, to do this operation.

The 30-in. (762 mm) lathes, however, were very well designed for this class of drilling and would have been especially superior for drilling the large hole. With both the work and the drill revolving, proper drill speeds could be secured and the time of operation shortened. In drilling the large hole only the work was revolved.

DRILLING THE 1 9/16-IN. (39.6875 mm) HOLE

A general view of the 24-in. (609.6 mm) heavy boring lathe with revolving steadyrest, with the fixture and recoil body in place and the drill in working position, is shown in the headpiece (Fig. 18). The spindle is driven by a 3-hp. (2.2371 kW) d.c. motor with pushbutton control and dynamic brake for stopping and starting and a field rheostat for speed regulation. The switch and the starting rheostat for the oil pump are shown mounted at the right.Fig_18

(no caption for this, but it is the headpiece and thus Fig. 18)


Fig. 19 shows the recoil held in the revolving fixture A which is clamped to the faceplate by the clamp B, while the other end revolves in the steadyrest C. The fixture is turned to the same diameter at both ends. The hole is-first drilled half way through from one end, then the fixture is turned end for end in the lathe without unclamping the recoil body and the other half of the hole drilled from the opposite end. The tools met within 1/64 in. (0.396875 mm) and often within a few thousandths.


Fig. 19. Details of holding fixture (Cyan overlay of letters and arrows added by me for clarity.  Also the "Cradle forging" lettering was added to show where the forging was in the machine- but that is the only other addition I made to any of the photos in this article)


The fixture is revolved at 70 rpm. in a counter-clockwise direction, viewed from the carriage. The recoil body is held by the clamps shown in Fig. 19. The revolving steadyrest C and the end of the fixture A are seen in detail in Fig. 20. The bushing F is held tightly against the end of the recoil body by the two screws shown. The oil-tube drill G is fastened to the hollow drill tube H and revolved in a clockwise direction viewed from the carriage. The drill is guided in the guide bushing E and in the stuffing-box J.


Fig. 20. How drill is driven and lubricated



Oil under pressure enters the stuffing-box J as shown and passes through the drill tube in the guide bushing E. The drill tube is smaller than the cutting diameter of the drill. This allows the oil to pass into the drilled hole and around the tube and thence along the oil clearance and over the end of the drill. As the chips are produced they are carried back through the drill tube by this oil.

The drills are shown in Fig. 21 and were made of Rex AA high-speed steel. The shank section was left soft and the land above hard. If the drill is not hard at the land, it will seize in the hole and twist off in use. It is necessary to give the oil clearance L the shape shown in order to prevent the wedging of fine chips along the edges.


Fig. 21. Some of the drills used


Another important point is the shape of the grooves for chip breaking shown at N. These grooves must have square sharp corners to properly split the chip into three parts. The shape and finish of the throat of the drill where the chips enter the hollow interior must be as shown and must be polished to prevent any clogging of the chips. The short drill shown at A is ready to be scrapped; it has drilled 102 holes. Figs. 22 and 23 show the method of revolving the drill tube H. The unit O is connected to the regular carriage by a forced bar and is fed forward along the bed by a feed screw inside the shears. This drags the carriage on which is mounted the 3-hp. (2.2371 kW) motor, controller and gearing as shown. The gearing is 10 to 1; the motor speed, 1,750, giving a drill speed of 175 r.p.m. This, added to the speed of the work, gives a total speed of 225 r.p.m. The maximum cutting speed of the drill is 80 ft. per minute (24.384 meters per minute). This drive was installed as a temporary expedient but proved entirely satisfactory. A cast-iron shearing pin, 5/16 in. (7.9375 mm) in diameter by 1 1/2 in. (38.1 mm) long, drives the drill tube. This pin is reduced in diameter to 1/4 in. (6.35 mm) by a neck 1 in. (25.4 mm) from the end and is easily sheared off in case the drill sticks.


 Fig. 22. Driving the drill


The three-cylinder pump, shown in Fig. 23, driven by a 3-hp. (2.2371 kW) motor, supplies the oil. The tank connections to the pump contain fine meshed screens to filter out the fine chips. This shows the arrangement of troughs and the way in which the outlet A must always discharge into B, no matter where the carriage may be. The oil used was Houghtons' refrigerant base, 7 gal. (26.4979 L); paraffin oil, 50 gallons (189.271 L). 


Fig. 23. Handling the drilling lubricant


The drilling data may be summed up as follows:




Drill size

1.3594 in.

34.52876 mm

Drilling speed (total)

225 rpm

Cutting speed

80 rpm

Feed per revolution of spindle

0.0035 in.

0.0889 mm

Feed per minute

0.787 in.

19.9898 mm

Idle time, loading, turning around, etc.

1 hr. 10 min.

Cutting time

2 hr. 50 min.

Total time

4 hr.

Number of operators required per machine


Holes per drill average




















In drilling 726 holes, six drills were broken. Four of these breakages were before the adoption of the aforementioned cast-iron safety pin and the other two were due to the drills being worn so short that they had insufficient taper clearance at the shoulder end.

The large hole in the recoil body is drilled from both ends and the operation is very much the same as with the 1 9/16-in. (39.6875 mm) hole, the main difference being that the work only revolves. Fig. 24 shows the drill and the work. The same type and size of machine is used as for the small hole. The fixture is the same type and the function and design of the drill, stuffing-box and bushings, size and type of pump and the oil used, are the same in both cases.


Fig. 24. Drilling the large hole


The same points must be observed in making this drill as in making the small one. The details follow:




Drill size

2.3906 in.

60.72124 mm

Drilling speed

80 rpm

Cutting speed

50 rpm

Feed per revolution

0.0014 in.

0.03556 mm

Feed per minute

0.112 in.

2.8448 mm

Idle time, loading, turning, etc.

1 hr. 10 min.

Cutting time

9 hr. 10 min.

Total time

10 hr. 20 min.

Operators per machine


-- Edited by AN5843 on Friday 13th of December 2019 11:08:40 AM



Status: Offline
Posts: 31

The second and final post of part II:


Both the boring and reaming of each hole are accomplished in one setting of the piece. The boring brings the holes parallel and in the correct position, and the reaming sizes give the holes the proper surface for lapping.

A modified French type of tool was used for boring, while for finish-reaming, a wood-packed reamer was developed by the Singer engineers for this purpose. Both these tools have the same bar and are pulled through the holes, one after the other. Two boring tools are first pulled through, each removing part of the stock left from drilling. The last boring tool drawn through leaves the hole straight and in the position required. The finish-reamer is then pulled through and the hole sized and left smooth. About 0.002 in. (0.0508 mm) is left for lapping and this amount is all that is required to remove the reamer marks.

Fig. 25 shows the small-hole boring and reaming fixture mounted on the stripped carriage of a 24-in. (609.6 mm) heavy-duty lathe. The recoil body is clamped in position by four brackets. The spindle nose driving the bar is screwed to the spindle and the tapered end of the bar is drawn in by two tapered keys.


Fig. 25. The drilling fixture used


Fig. 26 shows the modified French boring tool for pull-boring. The body R is made of Rex AA high-speed tool steel. The chip is produced at the cutting edge and washed by the oil flow through the slot in front of the cutting edge and out through the interior of the hollow body. The fiber washers S are fitted to prevent the chips reaching the bronze bushing T. The bushing T is ground about 0.0005 to 0.0007 in. (0.0127 to 0.01778 mm) smaller than the cutting diameter of the reamer and is free to turn on the body R. When the tool is revolved and pulled through by the bar, the body R revolves within the bushing. The fiber bushings S are ground the same diameter as the cutting tool.

The bushing T is made of forged bronze. The tool is fastened to the solid boring bar by the keys shown at K. This tool differs from the French cutter which is fastened to the boring bar by a female instead of a male taper. This was expensive to grind inside and not easily removed from the bar. The French tool is also without relief on the outside, back of the cutting edges, and also in the space which in the Singer reamer is occupied by the fiber collar. This caused the reamer to seize in the hole and break off. The Singer tool is backed off to a line as is any reamer and breakage entirely ceased after this type was introduced. The life of this reamer averaged 32 holes. Two of these tools are pulled through to straighten and position the hole. (The French tool is likely the boring tool used by the French to make their own Canon de 75s using the original manual method.  Apparently it was badly designed if this article is anything to go by.)


Fig. 26. Some of the drilling tools

Fig. 27 shows the small reamer entered in the starting bushing at the beginning of its travel through the recoil body. The steel starting bushing is 0.0005 in. (0.0127 mm) above the size of the bronze bushing in the reamer and gives it a straight start. While working, the oil and chips come out of this bushing. In case the reamer is reduced in diameter by grinding, smaller bushings are furnished below the maximum size, in steps of 0.002 in. (0.0508 mm). Fig. 28 shows the coming-out end of the fixture. Bushing V is bronze. The oil is pumped in at this bushing as shown and the bushing is held tightly against the end of the recoil body. 


Fig. 27. Pulling the boring bar into the work



Fig. 28. Boring bar just leaving the work

Details of the reaming operations are:




First tool size

1.457 in.

37.0078 mm

Second tool size

1.552 in.

39.4208 mm

Revolutions per minute


Feed per revolution

0.014 in.

0.3556 mm

Reaming time

1 hr. 50 min.

Loading, changing, etc.

25 min.

Total time

2 hr. 15 min.

The-fixture-reaming is done in the same fixture as the boring, without removing the recoil body. The last boring tool is removed from the bar and the bar pushed through the hole until it protrudes from the rear. The wood finishing reamer is then keyed to the end of the bar and pulled through. No starting bushing is used, as the wood follows the bored hole, this wood being turned 0.007 in. (0.1778 mm) larger than the hole is bored.

The finishing-reamer is shown in Fig. 29. The body is made of chrome-nickel steel and the blades of a fine carbon finishing steel. Colonial No. 7 and Bohlers Gold Label proved well adapted for this purpose. After being machined and hardened the blades are lipped as shown in Fig. 30. The blades are then fastened to the bodies and packed out to allow for grinding to the correct diameter.


Fig. 29. The boring tool



Fig. 30. Grinding the cutting lip

Fig. 31 shows the blades in position, being circular-ground to correct diameter and tapered. The angle of this taper is 3 deg., followed by a short taper of 1 1/2 deg., 1/8 in. (3.175 mm) long. As the reamer is pulled shank first through the hole, the taper occurs on the end of the blade toward the shank. Fig. 32 shows a reamer being backed off on a Cincinnati cutter grinding machine No. 1 1/2. The edge is backed up to the circular grinding line. Fig. 33 shows the reamer being stoned. The soft copper bar in the toolpost is used as an indicator and the two blades stoned until they both register exactly. The point where the taper meets the straight part of the blade is very carefully stoned.


Fig. 31. Grinding outside of boring tool



Fig. 32. Grinding clearance on boring tool



Fig. 33. Honing and testing boring tool

In Fig. 34 is shown the wood packing being turned. The wood is hard maple, soaked in hot cutting oil until the oil has permeated the wood thoroughly. It is left in the oil until shortly before the reamer is to be used and the turning is done just before using. That part of the wood which extends back along the blades is turned 0.007 in. (0.1778 mm) larger than the blade and is tapered at the shank to facilitate entry. This supports the blades as the wood leaves the hole on finishing. The oil flows back around the bar and washes the chips away as they are produced, as with the boring tools. The wood is shimmed out several times with veneer and returned. The blades are set out and re-ground for every cradle.


Fig. 34. Turning the wood packing

After some experimenting it was found advisable to pull two finishing reamers through the holes. This was mainly to counteract a tendency to taper the hole as the reamer finished cutting and the wood got away from the supporting walls of the hole in the cradle. Care must be taken to break the comers of the entering end of the hole to prevent shearing of the wood. When the wood is about right, a squealing noise is often heard as it revolves.

The details of this reaming are shown in the accompanying table.

DETAILS OF REAMING (for the small hole)




First reamer

Size of first finish-reamer

1.565 in.

39.751 mm

Revolutions per minute


Feed per revolution

0.0685 in.

1.7399 mm

Second reamer

Size of second reamer

1.572 in.

39.9288 mm

Revolutions per minute


Feed per revolution

0.050 in.

1.27 mm

Time changing from French tool to first finish-reamer, including gear changing

20 min.

Reaming time

17 min.

Change to finish-reamer

12 min.

Reaming time

45 min.

Removing work

15 min.

Total time two reamers

109 min.

(Final hole size after lapping is thus 1.574 in. (39.9796 mm).)

-- Edited by AN5843 on Friday 20th of December 2019 07:23:31 AM

-- Edited by AN5843 on Friday 20th of December 2019 07:31:55 AM



Status: Offline
Posts: 304

Thanks for posting this extremely detailed information about the French 75mm mle/97 recoil

Ralph Lovett


Status: Offline
Posts: 31

You're welcome, I hope it is useful to anyone who really wants to know about that gun and the mass production achievement the recoil system represents.

And with that, on to the first post of Part III:

Unusual Methods of Securing Extreme Accuracy—III


Consulting Engineer 

The third installment continues the boring and reaming operations on the two main holes in the cradle forging and then takes up the machining of the trunnions with a description of the unique fixtures and special machines used. 

(Part II appeared in our April 29 issue.) 


The boring of the large hole differs from the small hole only in detail. Fig. 35 shows the boring tools. The body X is soft steel, Y is Rex AA high-speed steel, and fiber ring R is used to stop the chips, instead of having a full-diameter section below the chip openings as the French tools have. These tools are very satisfactory. Care should be taken to see that the slots in the cutting tool line up with the slots in the body to give a clear passage for the chips.


Fig. 35. Tools for large hole

Fig. 36 shows the circular-grinding of the tool and bronze bushing. The bushing is ground 0.0005 in. (0.0127 mm) below the cutter size. After circular-grinding the cutting edge is backed off to a line. The small-hole boring tools are ground in the same manner. A fixture similar to the one used for drilling is used for boring and reaming the large hole. One end is clamped in the revolving steadyrest on a 24-in. (609.6 mm) heavy-duty lathe as shown in Fig. 28, and the other end in a special bracket fastened on the stripped carriage and bored in position, as in Fig. 37. Fixtures of the type used on the small hole were found satisfactory in this case. Fig. 38 shows the boring tool leaving the work. Fig. 39 shows the method of fastening the large bar to the spindle of the lathe. The bar screws into the collar A which has a center fitting the lathe spindle. This collar is then held in the spindle by the capscrews shown. The large finish-reamer is like the small one. 


Fig. 36. Grinding the boring tool



Fig. 37. Boring tool entering work



Fig. 38. Boring bar and oil connection



Fig. 39. How the bar is driven

The large-hole operation details, including removing the finished work from the fixture, are shown in the accompanying table.


 1st Bore2nd Bore1st Ream2nd Ream
Size2.476 in.62.8904 mm2.572 in.65.3288 mm2.586 in.65.6844 mm2.596 1/2 in.65.9511 mm
Revolutions per minute909055 to 6030
Feed per revolution0.014 in.0.3556 mm0.014 in.0.3556 mm0.0685 in.1.7399 mm0.050 in.1.27 mm
Time of cutting55 min55 min.17 min.45 min.
Setting up, etc.20 min.20 min.25 min.45 min.
Total time70 min.70 min.42 min.74 min.

(Final hole size after lapping is thus 2.598 ½ in. (66.0019 mm).)

This method of reaming is very satisfactory for this class of work. The holes come very straight and parallel, true to size and remarkably smooth; 0.002 in. (0.0508 mm) is left for lapping and polishing. After this method of drilling and boring was started, not a single cradle out of the 800 was spoiled in this operation. As little as 0.002 in. (0.0508 mm) has been removed from a hole with the wood reamer. The three cylinder oil pumps used deliver plenty of oil at 80-lb. pressure (551.581 kPa). Care should be taken to screen or otherwise remove the fine chips. The boring operations are covered by operations 16, 17, 18 and 19. The lapping was not done immediately after boring but was made one of the final operations. This was due to the fact that it is of extreme importance to preserve the final finish of the holes and consequently reduce the handling of the piece after lapping to an absolute minimum.

Operation 22 shows the milling of the grooves and surfaces for attaching the cover plates. These grooves were on both sides of the piece, but on the side shown in operation 22, the grooves extend the entire length of the piece, whereas on the other side these grooves were interrupted. Operation 22 was done on a horizontal milling machine using two gangs of cutters, the pieces being set at the proper angle as clearly shown in Fig. 40.


Fig. 40. Gang milling for side plates.

And that's it for the first post of Part III, this part has a lot of photos, so it will probably be split into 3 posts, maybe 4.  The second post will cover the milling of the trunnions, and the third will probably cover everything else.



Status: Offline
Posts: 31

The second post of Part III:

The next operation was the rough-turning and drilling of the trunnions. At this stage the trunnions were square blocks or bosses and it was deemed advisable to remove the large amount of metal still left after milling so as not to have to remove much metal at the finish-turning. This roughing operation was done on a high-duty drilling machine, the table having been removed and a special fixture substituted. This was really a hollow-milling operation, the hollow-mill being clearly indicated in Fig. 41. It will be noticed that the fixture can rotate on its axis so as to present both trunnions to the hollow mill. A number of other tools such as drills, reamers, taper reamers, etc., completed the equipment for this roughing operation and are shown in Fig. 42.


Fig. 41. Hollow-milling the trunnion



Fig. 42. Some of the trunnion tools

Skipping a few of the operations, we now come to the finish-turning of the trunnions. It was considered to be of extreme importance to have the trunnions exactly in line with each other. The peculiar shape of the piece and its great length make it very difficult to swing the piece around the center line of the trunnion on a lathe or vertical boring-mill. Besides, such a procedure would require two settings of the piece for tuning the two trunnions. Then, too, the method of hollow-milling both trunnions on a two-spindle machine would not insure perfect alignment of the two trunnions. For these reasons it was decided to build a special machine which consisted principally of a large pulley about 74 in. (1879.6 mm) in diameter as in Fig. 43. The pulley was made in halves, with the journals for bearings of the pulley entirely in one of the halves, the other half being merely a cover. Fig. 43 shows the cover as the upper part of the pulley.

In this position, the upper half can be removed by unclamping the swinging bolts which are clearly shown. The lower half is arranged as a receiving jig for the cradle as in Fig. 44. There are hardened bearing spots, adjusting screws, clamps, etc. A piece of work is placed on this lever half of the pulley and fastened, after which the upper half is put in place and the driving belt is thrown over the pulley. The machine is driven by a 7 1/2-hp. (5.59275 kW) motor with push-button control and powerful dynamic brake. There is, in addition, a foot brake of which the treadle is shown in the illustration. The motor is larger than is required for the running of the machine, but this amount of power is required for the starting up of the machine.

The trunnions of the pulley are hollow. On each side of the pulley there is a cast-iron stand on which is mounted a compound rest of an 18-in. (457.2 mm) engine lathe with a four-position tool block as seen in Fig. 45. After the piece is put in place, a smooth brass cover is put over the trunnion of the piece and inside of the journal of the pulley. This is for the sake of safety so that no projecting parts of the mechanism shall touch the operator or the tools and also to provide a non-changing field for the eye of the operator. An idler puts the proper tension on the belt and permits the belt being thrown off when lifting out a piece.


Fig. 43. Special machine for turning and boring trunnion.



Fig. 44. How the cradle fits in the machine. (This is the same machine as in the previous figure, seen from the front with the upper half removed; those bolts sticking up out of the sides hold the 2 halves together.)



Fig. 45. Boring the trunnion. (Again, this is the exact same machine as in the previous 2 figures, but from the right side instead of the left, and focused on the open milling section at the center of it.)


As there is a stand and compound rest at each of the two trunnions of the pulley, it is possible to have two operators work simultaneously. It is plain, of course, that there was no possibility of getting the trunnions of the piece out of line with each other. The only thing requiring attention was to have the axis of the piece at right angles to the axis of the pulley and this, being once established, would so remain during the life of the machine.

The rest of Part III will be split into 2 parts, just to keep the number of images per post down- most of these are just miscellaneous operations that don't fit into any specific group.



Status: Offline
Posts: 31

The third post of Part III:

Operation 24 and 24A are the milling of the interrupted slots. These required several settings and the use of sliding fixtures, one being shown in Fig. 46. The fixture shifts so that the hole B comes under the index pin A. Operation 25 was the fitting of the cover plates to the sides of the cradle. Operations 26 and 27 were the drilling, counterboring, tapping, and the fitting of screws for these cover plates.


Fig. 46. One of the small milling operations.


Operations 29 and 29 A were the planing out of the insides of the undercut portion of the cradle, 1/32- in. (0.79375 mm) being allowed there for finish. Fig. 47 shows the piece in position on the planing machine and one of the tools used. Operation 30 is the final finish-milling of the inside of the forging and confines itself to the lips and what is called the nerve guide. This finish-milling of the lips would not have been necessary if the original cutters had been entirely correct.


Fig. 47. Finish-planing the inside



Status: Offline
Posts: 31

The fourth and final post of Part III:

Operation 31 is the finish-turning of the trunnions. Operations 33, 34, 35, 35A, 36 and 37 are all minor milling operations shown in Figs. 48, 49 and 50. These give a good idea of the fixtures, the cutters and the gages used for determining the different cuts. Many of these cuts were at various angles, some were in the nature of an arc, others required peculiarly shaped covers, but none of them presented any unusual difficulties. The drilling, counterboring, etc., of the various holes were also ordinary operations, but some of the jigs and tools used were of interest. Fig. 51 shows the fixtures used for holding the piece vertically. Fig. 52 shows a simple drilling fixture which utilizes the trunnion as a locating point. Fig. 53 is a tapping fixture which was found very convenient.


Fig. 48. Milling end, and showing gages used



Fig. 49. Milling fixture for end, showing gages used for setting cutters



Fig. 50. Another end-milling job



Fig. 51. Drilling small holes in the end



Fig. 52. Locating side holes from trunnion



Fig. 53. A convenient tapping fixture

Parallel to the axis of the piece, and on each side of the center guide, there is a taper track with a very small taper per inch, which runs into a short piece of straight track. This straight part is the lower part of the track, very much like a street running level for a certain distance and then suddenly going up hill. The changing from the straight to the taper part was sudden, making a straight line across where they met. This required a certain amount of hand-finishing at the junction of these two parts and, in order to reduce this hand-finishing to a minimum, the following procedure was adopted:

The operation was done on a No. 5 milling machine with vertical attachments and lengthened table so as to give 60-in. (1524 mm) table travel. First, the straight part of the track was milled out with an end-mill to within 1/64 in. (0.396875 mm) of its final depth. Then the jig was tilted to the proper angle and the bevel part of the track was finished. As the straight part was not made to its full depth, it was possible to mill the straight part further than it should go and yet not touch the plane of the finished bevel part, so that, when this bevel part was finished, there was only a little corner left, 1/64 in. (0.396875 mm) in height. This had to be removed by hand. Operations 40, 40A, 40B, and 41, 41A, 41B, and 41C were simple, presenting no difficulty.

And that's the end of Part III.  There are surprisingly few dimensions mentioned in this section, but many figures.  The final (but most important) section has only 9 figures, so it should fit into 2 posts.



Status: Offline
Posts: 31

And moving on to the first post of Part IV, the final and most important part of the Canon de 75's recoil system:

Unusual Methods of Securing Extreme Accuracy—IV


Consulting Engineer 

The concluding installment describes the most important and difficult operation on the cradle forging, the lapping to finish size. The article ends with a brief account of the high-pressure hydraulic test to which each finished part is subjected. 

(Part III appeared in our May 18 issue.) 

The real trouble began at operation 42 which was the lapping of the small cylinder. Two thousandths of an inch had been allowed for this lap ping and proved to be sufficient to remove the tool marks of the finish-reaming. It was originally thought that it would not be possible to do this lapping entirely in a mechanical way and that a considerable time would be required to instruct and train men in the art of lapping. However, it was not thought practical to attempt to train men to a point where, by the mere skill of hands and eye, they would be able to produce both the extreme accuracy required and the beautiful finish which was said to be necessary.

The following method was adopted and proved to be entirely satisfactory for producing holes of the proper finish and of much higher accuracy than required in the specifications. It is believed that the success with the recoil mechanism as manufactured by the Singer Manufacturing Co. was largely, if not entirely, due to this method of lapping.

(Because of these statements and the previous other sources, this is likely the aspect of production that required the critical hand-polishing to incredibly high tolerances, with the precision of jewelers.  The lapping method is, therefore, the operation that was believed to be impossible, and the critical innovation that made the mass production of the recoil system possible.)

The lapping was divided into two processes, lapping and polishing. The lapping was for the purpose of producing a hole of proper size and roundness, the polishing for the producing of the mirror-like finish. The amount of metal removed in the operation of polishing was not measurable and was probably more in the nature of a burnishing operation without much abrasive action.

Two styles of machines were used for these operations, though the style which was used for the lapping would have been suitable also for polishing. However, the fact that some machines of the second style were immediately available, and that the first style was not absolutely required for polishing, was the deciding factor in making this division. The main difference between the two machines consisted in the fact that the piece to be lapped was held in the first style of machine in an indexing fixture, and in the second style of machine was held stationary. The first style machine is shown in Fig. 54 which shows the entire machine, and in Fig. 55 which shows the feed mechanism and the holding fixture. The feed mechanism was disconnected for short-stroke lapping.


Fig. 54. The machine for lapping holes



Fig. 55. Details of feed mechanism (In the actual article there is an error; the captions for Figs. 54 and 55 are switched.)

The machine consisted of a bed A on which was mounted the fixture B, in which the holding fixture C could revolve on trunnions D. At the outer end of the machine a motor E was mounted. This was a variable-speed motor of 10 hp. (7.457 kW) capacity, running from 600 to 1,800 r.p.m. but of which only the lower speed was used. The motor was reversing and the reversing mechanism F operated by the dogs H and the tappet G, Fig. 54, would reverse the motor in a small fraction of a second, so that it was even possible to use the machine on strokes as short as 18 in (457.2 mm). This was an extremely exacting duty of motor and controlling mechanism, but this part of the apparatus, furnished by the General Electric Co., stood up well and did not give any serious trouble. The motor shaft was directly connected to a screw I which had 3 1/2-in. (88.9 mm) lead, double thread, and would give a lapping speed of 175 ft. per minute (53.34 meters per minute). The lapping was not done, as is usually the case, by rotating but by reciprocating the lap in such a way that about half the length of the lap would project beyond the end of the piece at the end of each stroke. After each stroke the piece would be rotated through a small angle, the rotating mechanism being shown in Fig. 55.


The feed was pneumatic, a feed dog tripping the inlet valve so that the air can push the piston and rack to the right. At the extreme right the piston closes the inlet and opens the exhaust. A constant air pressure on the right side of the piston returns it to its position at the left, but before the end of the stroke the rack closes the exhaust valve to form a cushion. The movement is transmitted through a one-way ratchet and wormwheel with gears, with teeth so calculated as to avoid uniform positioning of the work at each stroke.

Fig. 56 shows the tools used for the lapping operation. A is the shell of the lapping head quite worn out; B and C are new shells for the large and small holes respectively. A is mounted on a head with taper wedges for expanding the shell by means of the nut D; E is the hollow bar to which the head is attached.


Fig. 56. The kind of laps used

The lapping compound consisting of No. 4 F carborundum and vaseline was forced through the hollow bar by means of a little force pump. It was squeezed through the small holes visible in the shell and carried by the right- and left-hand spiral grooves of the shell to the work. As the piece and not the bar revolved, or rather indexed, every part of the hole was subjected to exactly the same action. The result was that the hole became round. The fact that all of the outside surface of the shell carried lapping compound made the action relatively easy.

After lapping, the cradle was washed in soda and hot water as shown in Fig. 57. It was then put on the second type of machine as shown in Fig. 58. In this machine the piece was held stationary, and the lapping bar reciprocated and was also indexed. The reciprocating motion was obtained by a planing-machine drive and the feed by means of ratchet pawl and dogs. In this machine the same style of lapping bar was used but with an entirely different lapping head. In this case the lapping head was made of aluminum and of the shape as shown in F and G respectively in Fig. 56. The stones H and I were placed in the recesses of the aluminum head, spread apart by the springs J and kept in position by a piece of twine before they were inserted in the hole. The stones used were made of the so-called "water-of-Ayre" stone. This stone is exceedingly soft and being absolutely free from grit, does not produce any scratching. Great care had to be taken, however, to wash all of the abrasive material out of the hole before going over to the polishing operation; a grain of carborundum left from the lapping might imbed itself in the water-of-Ayre stone and scratch the surface. The reason why it was not so important to have the work indexed during the polishing lies in the fact that this polishing did not remove a measurable amount of metal.


Fig. 57. Washing holes after lapping



Fig. 58. Machine for polishing the holes

In the original method of lapping, but which was not followed by the Singer company, a lapping head was used similar to the one used for polishing. Carborundum stones were used as abrasive material with thin oil as lubricant and the lapping bar was indexed while the work was held stationary. This was satisfactory so long as the two stones were in a vertical position, but when the stones were horizontal the weight of the head and bar would cause the stones to rub in the hole with a wedging action and gradually cause the center of the hole to drop. In other words they had a tendency to produce an oval instead of a round hole. In the machine which was originally designed for this purpose, the lapping bar was supported close to the work. This in the opinion of the writer may be classed among many other good intentions, and led in the same direction as good intentions are generally said to lead. This support caused the lapping bar to overhang very little at one end of the stroke and very much more at the other end, so that the tendency to wedge, and to cause an oval hole, was not the same during the entire length of the hole. The result was that the hole was not only out of round, but that it did not have the same section throughout.

The pieces which have to move into these holes have to be packed with extreme care and this packing must conform to the section of the hole. If the hole was oval but of uniform section, the packing might adjust itself gradually to this oval form and thereafter work satisfactorily. But if the hole is of non-uniform section, then the packing must assume a different shape every time it moves along the inside of the hole, and this would certainly lead to failure in the function of the mechanism. It was for this reason that it was thought absolutely essential to have the work rotate and to have the bar entirely free and without support.

And that is the most critical part of the whole operation.  The second post of Part IV, and the last post of the entire article, will cover some final minor operations, but this is what I found to be the most interesting and valuable part to take away from this article.



Status: Offline
Posts: 31

And the final post, the second post of Part IV:

In addition to turning trunnions on the outside, they are also bored and threaded with the aid of special attachments and devices. One of these is shown in Fig. 59, where the trunnion is being reamed with the taper reamer shown. The fixture for holding the work by the outside of the trunnion is plainly shown, as well as the long bushing which guides the reamer into its correct position. Fig. 60 shows the taper reamers, as well as the tools for counterboring for the thread.


Fig. 59. Reaming the trunnion



Fig. 60. Tools for reaming trunnion

In Fig. 61 is another special machine built to mill the threads in the end of the trunnion, and also to line-ream the brackets which carry the sighting mechanism after it has been assembled in place. The thread-milling is done by the nearest spindle, while the one in the background reams the two holes to insure their correct alignment. After the cradle is completed, the holes, which are threaded at the ends, as shown in Fig. 62, must be plugged and tested to be sure the steel is of the proper strength. The kind of connection used is shown in Fig. 62, where the small hole is being tested. This is tested to a pressure of 5,100 lb. per square inch (35163.26 kPa).


Fig. 61. Milling trunnion thread and drilling



Fig. 62. Testing the forgings at high pressure


And that's the article on manufacturing the Canon de 75's recoil system in the US using mass production techniques.



Status: Offline
Posts: 304

I really only know of one person that has recently rebuilt a French 75mm mle/97 recoil system for live fire.  That is Matt Switlik.  This is a link to a YouTube interview with him about the French 75mm:

This YouTube link shows the 75mm mle/97 in Matt's collection live fire with projectile and recoil:

By the way, I have a French 75mm mle/97 but without a functioning recoil system. 


Ralph Lovett


-- Edited by Ralph Lovett on Wednesday 25th of December 2019 04:04:28 PM

Ralph Lovett


Status: Offline
Posts: 31

I found another series of articles by De Leeuw in the magazine Modern Machine Shop, with more details about how the Canon de 75 recoil system was manufactured.  It also has information about the methods used by the Rock Island Arsenal and the French Puteaux Arsenal.

 The first article is in the January 1937 issue of Modern Machine Shop (Volume 9 Issue 8), on Page 102:

"Inspecting Bores in 75 mm. Gun Recoil Mechanism


READING the article, “Inspecting Hollow Cylindrical Bodies with a Movie Camera,” by Rene W. P. Leonhardt in the November issue of MODERN MACHINE SHOP, I recalled a method I devised for examining bores in the main member of the recoil mechanism of the 75 mm. French guns.

There were two of these bores, both 63 in. (1600 mm) long, one with a diameter of practically 1 9/16 in. (39.6875 mm) and the other 2 1/2 in. (63.5 mm) Besides being held to very close limits as to roundness, straightness, and size, the holes had to be finished to a mirror polish.

 The method I employed to examine whether the finish was that of a true mirror was devised after I had first used a periscope but found this method entirely too slow and not entirely sure. Too much dependence was placed on the human element. The method I used, therefore, after the first two or three pieces had been inspected was the following:

A brass cap was inserted at each end of the bore to be inspected. The cap at one end consisted of a short section of tubing, which fitted the hole, to which a disc of very thin brass was attached, closing the end. A fine pin-hole was drilled in the center of the brass cap. The cap at the other end of the bore was of similar construction, except that the end metal did not need to be so thin, and that the central hole was about 11/32 in. (8.73125 mm) in diameter.

When examining the bore, both caps were put in position and a concentrated light was placed behind the fine pin-hole. Upon applying the eye to the hole in the other cap, one would see a bright, narrow ring of light halfway down the length of the bore, and a fainter ring of light at one-fourth and three-fourths of the distance. Otherwise no light would be seen, if the bore had a true mirror finish. If the finish were not correct, other light spots would be seen. This method would enable the inspector to pass on the finish in one minute or less.

It would be possible to arrange things so that one would see more than three light rings. It is even possible to see only one. If the light source is sufficiently concentrated and is at just the proper distance from the small pin-hole, then it may very well be that the only place from which that light can be reflected and strike the eye is just halfway. But this would require very accurate placing of the concentrated light.

If one is satisfied with the three rings as described above, it is not necessary to locate the concentrated light quite so accurately. There is quite some leeway possible.

This method is not recommended for examining bores in general. It only applies if the hole is finished to a high polish."


The second article is in the June 1937 issue of Modern Machine Shop (Volume 10 Issue 1), on Page 70:

It is also found in Google Books:

"Method of Polishing Holes in 75 mm. Recuperator Frames


Consulting Engineer

IN THE January, 1937, issue of I “Modern Machine Shop” I described a quick method to determine whether the surface of the bores in the frame of the recoil mechanism was a true mirror surface. The method was interesting on account of the simplicity and speed with which these bores could be inspected. The method of generating these mirror-like surfaces is interesting, too, because it departed from the well established method the French were using and had been using for a number of years.

The French arsenals used the honing method, that is, they did the polishing with abrasive stones. These stones were of approximately the same curvature as the bore to be polished. They were held in a metal head and pressed outward by means of springs. In short, the construction of the honing tools resembled in all essentials the honing heads now used for the finishing of automobile cylinders.

These heads were attached to the end of a long bar—long, that is, as compared to the diameter of the holes, for these holes were 63 in. (1600 mm) long and the largest was only about 2 1/2 in. (63.5 mm) in diameter. The bar was attached to a cross-head, and this in turn to a connecting rod that received its motion from a crank with sufficient stroke to cause the stones to rub along the entire length of the bore. When the crank had made a complete turn, the bar with the head would be indexed through a certain angle. Lubricant was pumped into the holes and the entire action was automatic until the hole was almost of the required size. After this point had been reached the operation became very laborious and required the greatest care and skill.

It would be found at the end of the automatic part of the operation that the holes were not entirely round. The vertical diameter would be little larger than the horizontal one. The workman would disconnect the indexing mechanism and change the position of the polishing head while the machine was making its strokes back and forward, until the hole was of the proper shape. Of course, it was extremely difficult to obtain the correct shape of the hole and at the same time reach the correct diameter. It should be kept in mind that the tolerance for the size of these hole was .0008 in. (0.02032 mm)

Besides correcting the holes, the workman was supposed to do something else, equally laborious. He was supposed to make the ends of the holes of a slightly larger diameter. The length of the enlarged part of the bore was only about two inches (50.8 mm). It would have been simple enough to do this if the machine had been capable of making a short stroke at the beginning and the end of the hole, but this was not the case. The machines had no provision for shifting the crank pins, but even if they had, this would not have helped. It would then have been possible to get a short stroke, but this stroke would have been in the middle of the length of travel while the requirements of the case called for a short stroke at each end. The workman had to provide the motive power as well as the control for this part of the operation.

We had complete information as to the way the French constructed the recoil mechanism, and could have followed their methods, but I felt that it would be impossible to train men to do such delicate work in the short time available. Besides, it took an average of 36 hours for one hole, and, at the rate we were supposed to construct these machines, we would have needed 90 polishing machines of the French type, even if the work were going on 24 hours a day.

One of the arsenals was also engaged in the work of producing these recoil mechanisms. They followed the French method with some variations. The polishing machines were furnished by an American machine tool manufacturer. They looked more like machines than the home-made French product, but had the same essential features. They used abrasive stones that could be indexed, and had one additional feature that caused an additional trouble. The bar that carried the polishing head was guided in a jig eye close to the entrance of the hole. The result of this construction was that the axis of the hole was slightly bent and had to be corrected. We can gather that this jig eye was made a feature of the machine on the general principle that jig eyes are good things, but a little thought should have shown the originator of this idea that the thing was very much out of place in this case.

At the beginning of the operation, there was just enough overhang of polishing head and bar to allow it to start into the hole. This overhang beyond the jig eye caused the bar to bend a very small amount.  However, when the polishing head was at the other end of the bore, the overhang and the consequent bending was much more, so that the head had the tendency to follow a curved path. In addition, this changing leverage of the weight of the polishing head did something else.

It was mentioned that the holes would be of oval shape if the French method were followed. This was caused by the fact that the polishing head was indexed.  When the head was in the position where one stone was at the top and the other one at the bottom, there was a slight difference in the pressures at the top and the bottom due to the fact that in the one case the pressure of the springs against the stones was hindered, and in the other case was helped, by the weight of the head and the bar. However, when the stones were at right angles to that position something else had to be considered. The weight of head and bar caused the stones to wedge in the hole. More material would be removed per stroke than in the vertical position and the head would drop, which caused the cross section of the hole to be oval. It is true that the amount of variation from the true circle was not great, but with a total tolerance as small as was the case here, even the smallest possible amount had to be considered. In any case, the fact that the hole had to be corrected for roundness when the French method was followed shows that this varying action of the head in its different positions was of practical importance.

If we now examine what the result must be of the method followed by the American arsenal, we find that the holes must be oval, that the shape of the oval—its eccentricity—is not the same at the beginning and the end of the hole, and that the axis of the hole is bent downward from beginning to end.

The importance of these factors was realized before we started the operations and even before we started the the design of the machines that were employed for this work.

In these machines the polishing head had no other movement than forward and backward. The workpiece itself was indexed at the end of a complete stroke. As a result, the action of the tools was the same on any part of the bore. There was no more tendency to wedge the tool into the bore at one point than at another. There was no jig eye, and so the sag of the head was the same along the entire length of the hole. Finally, the machine was so constructed that a short stroke could be had at any point of that length. At first it was decided to use the same kind of polishing head as the French used, but even before the operations were started, this idea was abandoned. It was decided to use the lapping method instead of honing.

The tool used for that purpose was a thin cast iron cylindrical shell. The shell was sawed in at both ends, six saw slots being provided at each end, extending two-thirds of the length of the tool and placed so that the slots at one end came between those of the other end. This shell was mounted on an expanding arbor so that a fractional turn of a nut would adjust the diameter to correspond to the gradually increasing size of the hole and also make up for the wear of the tool itself.

Forty-five degree shallow helical grooves were milled around the shell. There were two such helices, one right and one left hand. At the intersection of these grooves, holes were drilled through the shell and there were holes and grooves in the arbor so that lubricant introduced into this arbor would flow out through the small holes at the intersections of the grooves. The head was fastened to a hollow bar and this bar, in its turn, was connected to the driving mechanism.

The driving mechanism consisted of a screw with large lead, 3 1/2 in. (88.9 mm), directly connected to the driving motor. The motor was of relatively low speed and was reversible. The control of the motor allowed for a reversal in a fraction of a single revolution, so that a stroke of as little as two inches was readily obtainable. Dogs placed along the path of the bar took care of the reversal and by placing them at the proper points a short stroke could be had anywhere along the length of the hole. These same dogs also took care of the tripping of the indexing mechanism. The machine was quite simple and took up a minimum of space. A pump provided an ample stream of lubricant charged with a fine abrasive.

The machine required no particular skill to operate. The nut of the expanding arbor had to be adjusted from time to time until the proper diameter was obtained. The stroke was then shortened for one end of the hole and later for the other end. A few of these short strokes were required for the desired amount of bell-mouthing. The total time required for one hole was from two and one- half to three hours. The amount of metal removed was from 0.0025 to 0.003 inch (0.0635 to 0.0762 mm). All holes showed an accuracy far greater than what was demanded. They all came to within 0.00025 in. (0.00635 mm), whereas the tolerance was 0.0008 in. (0.02032 mm) When these holes were finished by this method it was found that one requirement was not met. The surface was dull, though very smooth.

The mirror finish was lacking. In order to obtain this finish, the lapping head was removed and the old honing head substituted. However, the original abrasive stones were not used. Instead, pieces of water-of-Ayre stone were used. These stones produced the desired finish in half an hour or less, and, although they changed the nature of the surface, they did not remove an amount of metal that could be measured with the instruments at hand. The total time from floor to floor for one hole was from three to three and a half hours, and this time was so much better than the best that could be obtained by the French method that no night work was required, though only twelve of these machines were built. More than sixteen hundred holes were treated this way, without a single failure."



Status: Offline
Posts: 36

Thanks a lot for the thread! I knew that the 75 was difficult to produce, but I couldn't imagine it was so bad!

For the context, in modern terms tolerance of 20 um for a hole of 63.5 mm (BTW, the number of digits in your metric conversions is excessive) would make it IT6 grade, which, AFAIK, is not used now in mechanical engineering for anything but gauges and certain details of measuring tools and devices. Perhaps only nuclear weapons may be made to similar and even stricter tolerances today but obviously the details are classified (in fact, the passage "it never refers to the 75 mm gun" made me giggle because it reminded me of articles on the Rosatom website which may describe production methods or technology optimization but never mention the purpose of the items produced).



Status: Offline
Posts: 36

The forum engine doesn't allow me to edit the post, but I wanted to add two more things:
1) Only other two artillery pieces I know to use the floating piston arrangement are the Soviet KS-19 100-mm AA gun (developed by Plant No. 8 in 1940s) and the US M777 155-mm howitzer (developed by Vickers in 1980s), I wonder what are the tolerances there? Apparently, there're just too much downsides in it, and the alternative designed by G. Canet for Schneider is just way more advantageous. As a side note, the terminology used with the M777 is quite different, for which refer to the patent US61259B1.
2) Attached is a diagram from a 1939 book on artillery materiel in Russian by David Kozlovskiy which was my best source about the design of the recoil system in question before I read your post:

-- Edited by ain92 on Wednesday 13th of July 2022 09:33:21 AM



Status: Offline
Posts: 31

Here are the 3rd and 4th articles about the Canon de 75 recoil system in Modern Machine Shop. These are the last ones in the series, to my knowledge.

I've moved the one illustration to a more appropriate part of this article, to avoid the confusion for which the 4th article below was written to correct.

The third article is in the August 1937 issue of Modern Machine Shop (Volume 10 Issue 3), on Page 66:

"Boring Long Holes in Frame of Recoil Mechanism of 75mm French Guns


IN a previous article I described the lapping of the two main holes in the main frame of the recuperator, or recoil mechanism, of the 75 mm French gun. The rapidity of that operation and the uniformly good results were made possible, in part at least, by the fact that the boring operations had made holes that were straight as to axis and of uniform and correct size. The amount of metal to be removed by lapping was from 0.0025 (0.0635 mm) to 0.003 (0.0732 mm) of an inch. This amount was kept uniform so that there were no appreciable differences in the time required for the lapping of the different frames.

These frames came to us as rough forgings, and the term “rough” applied with peculiar significance to them. When the original estimate was made as to cost and time required, the expectation was that these forgings would weigh about 950 lbs. (430.913 kg). As a matter of fact, they weighed 1350 lbs. (612.3497 kg), and, at that, some of them lacked sufficient stock in some places and consequently could not be worked up. Many others came so close to being useless that the greatest care had to be taken in the lay-out. Some of these forgings were bent, others had much stock on one side and not enough on the other one. The problem of lay-out was so serious that it was decided to build a special machine for that purpose. 

In this machine the forging was laid in a cradle which could be moved up and down in the direction of the axis of the forging; that is, it could be rocked in that direction. It could also be rocked in the other two main directions. This cradle was located in a bed, a lathe bed being used for that purpose. The ends of the cradle were turned as shown in the drawing, the ends being approximately 24 in. (609.6 mm) diameter. One end rested in a split cast iron ring attached to the face-plate of the lathe; the other end in a similar cast iron ring held in the steadyrest. 

When the cradle had been rocked to such a position that there was stock for the various operations at all points, it was locked in position by clamps furnished in the machine for that purpose. If it was found impossible to adjust the cradle so that stock would be available for removal on all sides, the forging was rejected. It may sound rather strange that this ever should be the case when one considers that the rough forging weighed 1350 lbs. (612.3497 kg) and that it weighed only 215 lbs. (97.5224 kg) when completely finished. Nevertheless, a number of forgings had to be rejected. 

There were two brackets, slideably mounted on the ways of the bed, one in front and one in the rear. Each of these brackets had a number of pins that could slide in them. Each pin was provided with a handle for manipulation. Some pins would slide horizontally and some vertically. The pins were supposed to strike the forgings when they were manoeuvered. If they were free to move without striking, this was proof that there was not enough stock at that point. Of course, the pins were located in the brackets to correspond to the cross section of the finished work. 

That a pin could be moved without striking was no proof that the forging was defective. It showed only that it could not be used if it were operated upon in the position in which it was lying in the machine. The cradle was then moved so as to correct its position in relation to that particular pin. Sliding the brackets inch by inch along the bed, and manipulating the different pins, it was possible to find in a very short time whether the forging could be used or not. 

Good forgings required about fifteen minutes for this examination. Bad ones took longer, for it would take considerable care and time before one could be sure as to their uselessness. At first, such forgings were laid aside for further examination without spending too much time on them for speed was an essential. As soon as there were enough good forgings so that operations could begin and continue for some time, the ones that were laid aside were subjected to a more careful examination.  

The usual method of lay-out is to scribe lines on the object to be machined. The method followed with these forgings was a different one. A special drill head was located at each end of the machine bed. Each head had two spindles. When the forging had been so located that in that position there would be stock for all operations, the cradle would be clamped and the drill heads set to work. Each head would drill two 1 1/4-in. (31.75 mm) holes in the end of the forging. These holes were the gage points for the first operation. This operation consisted in planing two parallel strips, and these strips were, thereafter, the gage points for all subsequent operations. 

 The entire operation, up to and including the drilling of the holes, required about twenty minutes. A few of the forgings had been laid out by hand before the machine was ready. This hand operation took about twenty four hours and the service of a very skillful and careful man. 

 As a matter of course, the forgings were completely roughed out before the boring of the holes began. These holes, it will be remembered, were 63 inches (1600.2 mm) long. The diameter of one was 1-9/16 in. (39.6875 mm) and that of the other 2 1/2 in. (63.5 mm). As a matter of fact, the dimensions were given in millimeters, but for a general consideration of the problem the dimensions as given here are sufficiently close. Fearing that the drilling of the holes might affect some of the dimensions in other places, some metal was left here and there for a final finish. However, it was found that this precaution was not necessary. Unimportant operations only were done after the boring of the holes. 

 After the preceding operations had been completed, there was only a thin wall left around the holes, so that the first drilling had to be accurate. Once the holes had departed from their proper directions, corrections were not practical unless the errors were small, 

 The French method consisted in drilling a hole half way through, then reversing the forging and drilling the other half. The method was also followed by Rock Island Arsenal, so it was reported to me, except that a duplex machine was used, drilling both halves of the holes in one operation. In either case, the drills revolved and the piece of work was stationary. 

 Our method was different, because it was realized that it is not possible to control the direction of a drill. As a matter of fact, it was found in the French arsenals that the two halves of a hole did not always meet and that, sometimes, a drill would break through the wall of a hole. This was to be expected, considering that there may be soft or hard spots in the forging, and that such spots would deflect the drill. Moreover, to make sure that there would be enough metal at all points for the subsequent reaming operations, more metal would have to be left for this reaming than would be necessary if one could be sure that the two halves of the hole would meet centrally. To overcome these difficulties, the following method was used.

The machines used for the drilling operations were lathes with long beds. Fixtures were built in which the forgings could be laid. There were, of course, the necessary devices to place and hold, the forgings in the proper position. Each fixture was turned at both ends and the center of such a turned part was the same as the center of the hole to be drilled when the forging was in the fixture. The face plate of the lathe was provided with a cast iron ring in which the turned part of the fixture would fit. The steady rest of the machine was also provided with a cast iron ring instead of the standard three jaws. Each lathe was further provided with an individually driven boring head. 


The principle underlying our method of boring is this, that the axis of the hole will be straight and that the cross section will be truly round if both piece and drill revolve. There is only one thing that may or may not be right when this method is employed. The hole may not be of even diameter. It may be tapered with the large diameter at the far end of the hole. 

So as to minimize the effect of this fault, only half the length of the hole was drilled at one setting, after which the fixture was turned end for end and the other half was drilled. Another reason why the holes were drilled in two operations was that, otherwise, the length of the lathe would become excessive. It was found invariably that the two halves of the hole met centrally and, though there was often some taper, this amounted to a few thousandths at the most. As there were three reaming operations to follow, this slight taper did not affect the final result. Incidentally, it may be mentioned here that this tendency to make a taper hole can be used to advantage in some cases, and that the writer has made some interesting applications of this fact. 

It is hardly necessary to point out that each of the two holes required its own fixture, for the drilled hole must be central with the spindle of the lathe. 

The drilling was done with a special twist drill. It was an oil drill, but no oil was introduced through the center of the drill, nor along the tube provided for that purpose; neither did the oil find its way out along the clearance surface of the drill. Quite the opposite took place. The oil was introduced along the clearance surface of the drill and oil and chips found their way out through the center. Oil was provided under a pressure of 300 lbs. to the square inch (2068.43 kPa). It would, perhaps, be better that the final pressure was of that amount, for so long as the oil could freely run out of the hole, the pressure was, of course, much less. However, that pressure would begin to build up as soon as the chips began to choke the hole in the drill, and it was not until that pressure had reached the amount stated that the chips were ejected.

Attached to the drill was a long hollow bar which was revolved by the mechanism of the boring head. A bushing was fitted in the end of the fixture and tight against the end of the work piece. Drill and boring bar fitted the hole in that bushing. It was through this bushing that the oil was introduced. As will be seen from this description, the oil entered the hole along the clearance surface of the drill. A short length of the hole was drilled before the bushing was placed tight against the work. As soon as this preparatory hole was long enough, the actual work of driling began. The oil would lubricate and cool the very ends of the cutting lips of the drill. The chips would find a passage through the center. 

So as to minimize the friction between chips and drill, the throat of the tool was nicely polished. As more and more chips tried to escape through the hollow drill and the long boring bar, there ensued a clogging up of the hollow bar and the pressure of the oil began to build up. When a pressure of about 300 lbs. (2068.43 kPa) was reached, a wad of chips was forcibly ejected and the process began anew. A feed of 1 1/8 in. (28.575 mm) per minute was obtained.

The smaller of the two holes was 1-9/16 in. (39.6875 mm) in diameter and the drilled hole 1 1/4 in. (31.75 mm) thus allowing ample stock for the various reaming operations. The reason why this large amount of stock was left was that, at the time tools and fixtures were designed and made, there was no absolute certainty that the drilled holes would be as perfect as they actually proved to be. The expectation was that these holes would be in line and of the proper cross section, but there was no previous experience to bank on. If the high expectations we had were not realized, then we would have to design and make new tools, unless there was enough stock left to meet all contingencies. This would mean loss of time and at that period time was of the first importance. It was decided, therefore, to allow enough stock after the drilling operations to be safe, even if the proposed method of drilling should not give as good results as we expected. 

There were three reaming operations. The tools for the first two were what are called “wood reamers.” Though not commonly used in the average machine shop, this type of reamer is well known in arsenals and other places where artillery is made. The cutting tool, itself, is a straight blade, mounted on a piece of wood. This wood support is of cylindrical shape. Part of this cylinder is cut away, so that the blade can be mounted on a flat surface. The wood part of the tool fits closely in the finished hole, so that the tool is its own guide. The blade acts as a boring and also as a scraping tool. Though the operation goes by the name of reaming, it is really a boring operation. 

The wood part of the reamer must be soaked in oil to facilitate its passage through the hole. It is a very simple matter to make oil penetrate into wood. All that is necessary is to let the wood lie in the oil for a sufficient length of time, but time is at a premium when there is war. It was not possible for us to let the wood soak for several months before using it, and so some other way had to be devised.

Several schemes were tried, one of which called for laying the pieces of wood in oil in a closed vessel in which compressed air could be admitted. This did not work out as desired. As a matter of fact, it was only tried because it was felt that nothing should be left untried. Another scheme, and one that was somewhat more successful, consisted in soaking the wood in hot oil. However, the depth of penetration was not sufficient.

The final, and successful scheme was the combination of hot oil and a partial vacuum. The vacuum was applied first, and after some time hot oil was admitted under ordinary pressure. The vacuum removed some of the occluded air and some moisture and made, so to say, room for the oil. It might be asked why an oily surface of the wood should not be sufficient. However, it should not be forgotten that the reamer must penetrate a long distance into the work and that it leaves some oil behind as it progresses.  There must be a reservoir of oil in the wood, for it does no good to pour oil in the hole while the cut is on. This may do some good to the cutting tool itself, but not to wood part, for this is squeezed tightly in the hole.

The reason why two reaming operations were necessary with wood reamers is that a considerable amount of metal had to be removed, and that the tool was liable to lose its size during its passage through the hole. As the second reamer removed only a small amount of stock, there could not be much wear of this tool and it might be expected that the hole would be of even size throughout after this second reaming operation. However there was a possibility that the axis of the hole was no longer a straight line due to the action of the wood reamers. It is true that the close fit of the wood part tends to guide the cutting blade, and that, in theory, the reamed hole should be as straight as the drill had made it, but, in reality, slight variations in the condition of the metal may deflect the tool, especially so as that tool is pushed through. Pushing it through means pressure in the long bar holding the reamer, and this means deflection of the bar and a consequent tendency to steer the tool in the wrong direction. It was, therefore, decided to follow up with a third reaming operation in which the reamer was pulled, and not pushed, through. This pulling has a tendency to straighten the hole. Of course, very little metal was removed in this last operation, so that there was no danger that the tools would wear appreciably during a single passage.

It is just possible that greater precautions were taken than were absolutely necessary, just as it may be possible that we might have started with a larger drill. However, the extra operations did not take much time and they made the final result sure. Measurements showed that, after the last reaming operation, the holes were just as accurate as after lapping; the lapping was merely for the purpose of getting the right finish.

The tolerance for each hole was 0.0008 in. (0.02032 mm) and for the distance between the two parallel holes 0.002 in. (0.0508 mm). This should explain why such care was necessary. We came well within the specified allowances. The diameter of the hole did not vary more than 0.00025 in. (0.00635 mm) from the basic figure."


The fourth and last article to my knowledge, is in the October 1937 issue of Modern Machine Shop (Volume 10 Issue 5), on Page 116:

"Additional Notes on Boring Long Holes in Frame of Recoil Mechanism of 75mm French Guns


IN my article on the boring of long holes in the frame of the recoil mechanism of the 75 mm. French guns (Page 66 August, 1937, issue Modern Machine Shop), mention was made of two cradles. In some way these cradles were mixed up. This is not as serious as the mixup of babies, which everyone must have read about in the daily papers; nevertheless it is confusing. 

One of these two cradles was meant for the lay-out machine and the other for the actual drilling of the holes. The sketch of one of these cradles shown in the first part of the article shows the essential elements of the drilling cradle, and the description of it at that point would make the reader think that it was used in the lay-out machine. Now, these two kinds of cradles were not at all interchangeable. 

In the lay-out machine, the work piece was so adjusted that all of the sliding rods would touch the piece— if this were possible at all. If it was not possible, then the forging was rejected, for it would not be possible to clean it up at all points. Once the piece was adjusted, something had to be done to make sure that, in the various succeeding operations, there should be no stock to be removed. For this purpose, the two holes at each end were drilled. These holes served as gage points for the first operation, which consisted in the planing of two strips. These strips then served as gage points for all further operations.

The reason why two holes, and not just one, were drilled at each end was that a single hole at each end would still leave the position for subsequent operations undetermined. The piece could be revolved around the axis if there were only one hole at one end. If there had been two holes, but at one end only, it would be possible to lay the forging down with one end too much to the right or left, or too far up or down. It would have been sufficient to have two holes in one and only one in the other, but this, too, would have led to some trouble, for it was sometimes necessary to turn the piece end for end in order to reach certain points with the tool. Besides, it took no more time to drill two holes at each end than it would take to drill one hole at one end and two at the other. 

These preparatory holes were drilled in the positions occupied by the final long holes. The fact that the forging had to be adjusted to some as yet unknown position when it was first laid in the lay-out machine precludes the idea that the cradle in which it was laid was centered in the faceplate of a lathe. The article speaks of a lathe bed used for the main frame of this lay-out machine, but this was only done because such a bed was available and eliminated the necessity of making a pattern and casting. The machine itself bore no further resemblance to a lathe. There was no faceplate, and, consequently, nothing could be centered with it.

However, the machine in which the drilling was done was a lathe. There was a headstock with a faceplate, and it was essential that the forging which, by that time, had undergone a number of operations, should be central with the spindle of the machine. The jig, or cradle, holding the forging for the drilling operations, was turned at both ends as the diagram showed. This permitted the jig to be placed exactly central with the faceplate and, therefore, with the spindle. 

It is hoped that the foregoing will clear up whatever confusion may have been caused in the mind of the reader by the misplacement of the diagram and its description in the previous issue of the magazine."



Status: Offline
Posts: 31

ain92 wrote:

Thanks a lot for the thread! I knew that the 75 was difficult to produce, but I couldn't imagine it was so bad!

For the context, in modern terms tolerance of 20 um for a hole of 63.5 mm (BTW, the number of digits in your metric conversions is excessive) would make it IT6 grade, which, AFAIK, is not used now in mechanical engineering for anything but gauges and certain details of measuring tools and devices. Perhaps only nuclear weapons may be made to similar and even stricter tolerances today but obviously the details are classified (in fact, the passage "it never refers to the 75 mm gun" made me giggle because it reminded me of articles on the Rosatom website which may describe production methods or technology optimization but never mention the purpose of the items produced).

Oh, I was just punching the numbers into Google to convert them, so that metric users have at least some values to work with- I have no idea how to convert the precision of a value from imperial to metric.

As for tolerances, there are quite a few parts today that do use tolerances this tight.  It has been noted that the "thou" (unit of a thousandth of an inch) was introduced in 1844 by Whitworth as a standard in machining, and this became a main standard by at least 1857.  At the same time, it was noted in at least one video that Richard F Moore gave industry an additional decimal place of accuracy, from founding the Moore Tool Company in 1924.  When checking Foundations of Mechanical Accuracy by J Robert Moore, it notes on Page 165 that the Jig Borers were accurate to about +/- 0.0002" (0.005 mm), and that in about 1950 they started to upgrade their standards for newer machines intended to be accurate to millionths of an inch (hundredths of a micrometer).  The original Jig Borers were considered a great advance, so it seems the extra decimal place was the ten-thousandth of an inch position.  So from about WW2, it seems the industry could regularly machine to the 0.0001" accuracy that the Canon de 75 needed.  And today, of course, much higher accuracy is common- as Foundations of Mechanical Accuracy states on Page 280, ten-thousandths of an inch dimensions started being specified in WW2, "split-tenths" of a thousandth of an inch were used in the post-war period, and when the book was written in the 1960's millionths of an inch specifications were common.  The best machine Moore made at the time, the subject of that book, could measure parts to an accuracy of 25 millionths of an inch in all 3 dimensions.  But in 1917, much less 1897, the Canon de 75 was a challenge.

There were however special cases before this, where this accuracy (or even greater) was attained before then.  As mentioned throughout Foundations of Mechanical Accuracy, diffraction gratings were made since the 1880's to whatever the tightest tolerance could be, regardless of cost.  In 1932, Henry Ford's book Moving Forward devoted a chapter to accuracy, which among other things noted the wrist pin hole in a Ford Model A had a tolerance of 0.0003", and some of the gauges used had a tolerance of 4 millionths of an inch.  One forum post showed some aircraft engine tolerances of a few fractions of a thousandth of an inch in WW2.  And most interestingly, the Antoinette aircraft engines of 1902-1910 were "made to the highest standards in a shop that boasted of tolerances down to .0004 in (.01 mm)."  This is especially important since they were French manufactured, just like the Canon de 75, and the above article by de Leeuw mentions that for the Canon de 75 "the construction of the honing tools resembled in all essentials the honing heads now used for the finishing of automobile cylinders."  These engines were built starting 5 years after the Canon de 75, but the shops most likely had similar tools and characteristics to the Puteaux arsenal.



Status: Offline
Posts: 31

ain92 wrote:

The forum engine doesn't allow me to edit the post, but I wanted to add two more things:
1) Only other two artillery pieces I know to use the floating piston arrangement are the Soviet KS-19 100-mm AA gun (developed by Plant No. 8 in 1940s) and the US M777 155-mm howitzer (developed by Vickers in 1980s), I wonder what are the tolerances there? Apparently, there're just too much downsides in it, and the alternative designed by G. Canet for Schneider is just way more advantageous. As a side note, the terminology used with the M777 is quite different, for which refer to the patent US61259B1.
2) Attached is a diagram from a 1939 book on artillery materiel in Russian by David Kozlovskiy which was my best source about the design of the recoil system in question before I read your post:

-- Edited by ain92 on Wednesday 13th of July 2022 09:33:21 AM

 The Recoil Systems Metric document has some overview of different kinds of recoil mechanisms, on Pages 3-9 to 3-15:

The Puteaux recoil mechanism has the following positive features: 

  1. Compact
  2. Lightweight
  3. Provision for a fluid index
  4. Possibility of variable recoil stroke
  5. One rod connection to the breech lug or to the front end of the cradle.

The Puteaux recoil mechanism has several characteristic limitations:

  1. Inadequate fluid reserve may allow the gun to fallout of battery at high elevation.
  2. Control rod is not positively attached to the gun; therefore, its correct position is not inherently assured.
  3. Repairs require special facilities and expert mechanics.
  4. Limited counterrecoil control is provided.


Desirable features of the St. Chamond recoil mechanism are 

  1. Variable recoil is provided at all elevations.
  2. It is compact.
  3. It is lightweight.

Undesirable features of the St. Chamond recoil mechanism are

  1. An inadequate fluid supply may permit the gun to fall out of battery at high elevation.
  2. No fluid index is provided.
  3. Repairs require special facilities and expert mechanics.


The particular advantages of the Filloux recoil mechanism are 

  1. Variable recoil to suit all angles of elevation is provided.
  2. Adequate counterrecoil buffing is provided.
  3. A fluid index is provided.

Some inherent disadvantages of the Filloux recoil mechanism are

  1. Inadequate fluid reserve may permit the gun to fall out of battery at high elevation.
  2. Repairs require special facilities and expert mechanics.
  3. The recoil and counterrecoil cylinders require separate filling.


The Schneider recoil mechanism has the following merits: 

  1. It provides adequate counterrecoil buffering.
  2. No floating piston is used.
  3. The control rod is secured to the gun to insure correct position.

The Schneider recoil mechanism has the following limitations:

  1. The recoil and counterrecoil cylinders require separate filling.
  2. No fluid index is included.


Of these, only the Schneider mechanism has a direct-contact recuperator, and as such it is the only one to not have the disadvantage of "Repairs require special facilities and expert mechanics."  But it also is the only mechanism to have the disadvantage of "No fluid index is included" which is normally attached to the floating piston and is obviously impossible without one.


Otherwise, the US has generally used floating pistons in most of its artillery after WW1.  Based on the sources I mentioned in a different thread, and the recoil section of the book Elements of Ordnance (specifically Pages 255-263 and 367), the US used hydrospring artillery prior to its entry into WW1.  The last generation of hydrospring guns- the 75 mm M1916, and the 3-inch M1918 AA gun, used a perforated sleeve that rotated for variable recoil.  

When the US entered the war, it adopted mostly French artillery (and some interim designs, but only until the intended new guns could be produced).  The Canon de 75 field gun used the Puteaux mechanism (as it was invented for that gun at the Puteaux Arsenal) as described above, as Elements of Ordnance and many technical manuals show.  The 155 Schneider C 17 S and 240 mm howitzers both used the Schneider mechanism, which was apparently used on all Schneider artillery in WW1 bigger than a field gun.  Both the Schneider 155 howitzer and the recoil mechanism of the 240 mm howitzers can be seen in a list of American Machinist articles I posted elsewhere, and they are described there.  The Schneider 155 howitzer recoil mechanism is also seen on Page 263 of Elements of Ordnance, and in the previous thread I linked to which mentions the "Handbook of Artillery" as a source.  The 155 GPF (M1918) gun used the Filloux mechanism, as Page 367 of Elements of Ordnance and the technical manual shows (on Pages 76-77, Figures 7-9). Since the GPF was invented by Filloux, it is likely the mechanism was invented for and first used on this gun.  The 8-inch Howitzer Mark VIII was British and is not mentioned in the above sources, but it used a direct-contact recuperator as will be described later.  At the same time, the US started to upgrade the 75 mm M1916 and 3-inch M1918 AA guns with hydropneumatic recoil systems.  This was the St Chamond mechanism, as described in the previous sources, and it was specifically invented by St Chamond for the 75 mm M1916 gun.  Alternately, though not mentioned in those sources, the Mark XI naval landing gun had a similar perforated-sleeve recoil system to the 75 mm M1916, and the 3-inch M1918.  But it used a hydropneumatic recuperator instead of a hydrospring one, as seen on Plate III of this book.

So by the 1920s, the US had almost all hydropneumatic artillery pieces, and all of them except the Mark XI landing gun, the Schneider 155 and 240 mm howitzers, and the British 8-inch Howitzer Mark VI used floating pistons.  It used all 4 of the above mechanisms described (Puteaux, St Chamond, Filloux, and Schneider) and then some, which is probably why later US manuals use these 4 as examples.


For the newer generation of artillery in WW2, the US seems to have used the Puteaux and Filloux mechanisms on almost everything, and both used floating pistons.  In particular, the Filloux Mechanism can be seen on:

The 155 mm Gun M1 and 8-inch Howitzer M1 - In TM 9-1350 on Pages 8-13 and TM 9-1350 on Pages 124-165 (these used the same carriage, and the 155 mm Gun was itself developed from the 155 GPF which used the Filloux mechanism)

The 4.5" Gun M1 and 155 mm Howitzer M1- in TM 9-331B on Pages 109-129, particularly Pages 118-120 (The 4.5" Gun and 155 Howitzer used the same carriage, and this refers to the self-propelled version, but it should have the same recoil system as the towed version described in TM 9-331A- unfortunately I couldn't find a free pdf of that online).

The Puteaux mechanism in turn can be seen on the smaller guns, namely:

The 105 mm Howitzer M1- in TM 9-1325, on Page 52

The 75 mm Pack Howitzer M1- in TM 209-319, on Pages 213-214


Finally, for more modern post-WW2 guns, the Recoil Systems Metric mentioned at the start of this post has a diagram on Page 4-6 of the 155 mm M198's recoil mechanism.  Known as the M45, the mechanism is based on the Puteaux mechanism, but it is heavily modified with features from the Filloux mechanism.  Specifically, the control rod has grooves in it and rotates to provide variable recoil like the Filloux mechanism, and it uses a separate replenisher like the Filloux instead of the Puteaux mechanism's fluid reserve.


For British artillery, hydrospring recoil was used on most of their early guns, as can be seen in the Wikimedia images classed under recoil mechanisms.  Upon searching for the documents listed as the sources for these diagrams, it is easy to find a bunch of other handbooks for British land artillery:

QF 15-pounder gun handbook- Page 15 (this was the first land-based artillery in British service with a full recoil system, and one of the first in the world, being the first produced version of the Ehrhardt Gun)

QF 18-pounder gun Mark 1 handbook- Page 13

BLC 15-pounder gun handbook- Page 22

BL 60-pounder gun Mark I Handbook- Pages 67-68

BL 6-inch Mark gun VII handbook- Pages 33 and 36

Just before and during the war a number of artillery systems were adopted with hydropneumatic recoil systems.  The ones that I can find diagrams for all use the same basic valve: the a rotating piston design known as either a Vavasseur-Ehrhardt valve, or a Krupp valve.  Some had floating pistons and some had direct-contact recuperators, as the diagrams in the sources below illustrate:

BL 9.2-inch Howitzer Marks I-II Handbook (US)- Pages 42, 44, and 47 (pages 54, 56, and 59 in this document).  Floating piston (also likely used on the BL 12-inch and BL 15-inch howitzers, since those were scaled-up versions of this howitzer), though built in lesser numbers than smaller guns

BL 6-inch 26 cwt Howitzer Mark 1 Handbook- Pages 51-53, 107-113 (Also found at this link, check Plates XV-XXI and Pages 41-43).  Direct-contact recuperator

BL 8-inch Howitzer Mark VI/VII/VIII Handbook- Pages 57-60, (Pages 47-50 in document), and also here with the plates included- pages 24-37.  Direct-contact recuperator (this gun was essentially a scaled-up 6-inch 26 cwt howitzer, with a recoil system that differs only in size and some minor details

BL 6-inch Mark XIX Gun Handbook- Pages 55-57, 116-120 (Pages 46-48 and Plates IX-XIII in manual).  Direct-contact recuperator (this gun used the same carriage as the BL 8-inch Mark VI, so the recoil system is identical)

BL 60-pounder Gun Mark II Handbook- Page 91.  Floating piston, though only built at end of war and entered service after the war ended (aside from that, the recoil system is very similar in function to the 6- and 8-inch guns and howitzers above, it only differs slightly in layout)


Based on this, at the end of WW1 almost all the British heavy artillery was similar in design, with only the big siege howitzers (the 9.2-, 12-, and 15-inch howitzers) having their own different design, though only those big howitzers and the 60-pounder Gun Mark II had floating pistons.  I can't find much information on the post-WW1 designs, though.


The French, as mentioned, used floating piston designs in the Canon de 75 and 155 GPF, and direct-contact recuperators in their Schneider artillery at least through the end of WW1, much like the US.  Russia used Schneider designs for all their modern artillery heavier than a field gun in WW1.  As such they used direct-contact Schneider recoil systems for virtually all their hydropneumatic recoil systems, and the USSR continued to use those designs in the modernized versions of these guns in WW2 and guns derived from them (up to the ML-20 and A-19).  Afterwards, it seems they still used the basic Schneider recoil mechanism until at least in the 1960's, with the T-12 antitank gun still being using it and a direct-contact recuperator.


I have no information on Austro-Hungarian artillery recoil systems, but a few pieces of information on German recoil mechanisms of WW2.  The document TM E9-325A covers the German 105 mm LeFH 18 howitzer, and is found here.  On Pages 20-24 (26-29 in the Archive document), it shows the recuperator to be a direct-contact design.  Similarly, TM E9-369A covers the German 88 mm Flak 18 family of AA guns, and is found here.  In Section II about the recoil mechanism, there are no cutaways, but by the description and external image the recuperator seems to be the same design as that on the LeFH 18.  So it seems Germany favored direct-contact recuperators in their artillery of WW2, likely WW1 as well.

That's what I could find on the subject of which recuperator designs were used in WW1 and later by various countries.



Status: Offline
Posts: 2281

The US used the Ehrhardt recoil mechanism in the 3inch M1902 field gun similar to the British 15 Pounder.

There were also a series of locally designed howitzers in 3.8, 4.7 and 6inch calibres which used hydrospring recoil mechanisms

and the 3.8 and 4.7inch guns which similarly used hydrospring recoil mechanisms.

Just about all of the locally designed guns were supplanted by British and French designs during WW1 mostly because US industry was 

making the better performing European guns before the US entered WW1. 

The Russians used a hydrospring recoil system in their 76.2mm M1902 field gun after the rubber ring recuperator of the 76.2mm M1900 proved

to be somewhat inadequate.  

I've never understood how the Mle 1897 gun didn't run into problems with the gas pressure and temperature in the recuperator rising to dangerous levels

during extended firing.  



Page 1 of 1  sorted by
Quick Reply

Please log in to post quick replies.

Tweet this page Post to Digg Post to

Create your own FREE Forum
Report Abuse
Powered by ActiveBoard