rexford

For bracketing by 75L48 APCBC vs T34 M43 at 1200m, following results were obtained (avg error of 16% in first round range estimate, single dispersion, firing by 20 guns vs 20 targets on first try): 4 of 20 first rounds hit target 8 of 16 second shot rounds hit target 3 of 8 third round shots hit target 3 of 5 fourth round shots hit target 18 of 20 targets hit by fourth round or less, as per German training requirements that a target be hit by fourth round from 1200 to 2000m. Average number of shots per target, 41 shots to kill 18 T34, for 2.3 shots per kill on average. For 1200m and 1500m bracketing exercises the results were 3.06 shots per kill at 1500m and 2.3 shots per kill at 1200m, darn close to the 1942 report for 75L43 against T34.

Please explain why it is so funny, and also detail how the Germans boresighted their guns.

The British firing tests which defined the 9.6 kg Pzgr APCBC round as the standard ammo for the 88mm L56 Flak gun were performed during February 1944, prior to D-Day. The report was dated November 1944. They could not have been from Home Guard units, and probably were obtained in Italy during 1943 (my guess). To the best of the British folks knowledge, 88mm L56 Flak guns were firing 9.6 kg Pzgr APCBC through early 1944.

The normal distribution curve, which is what the Germans and British used for dispersion results, can be defined by a mid-point and a standard deviation. The standard deviation for German dispersion data equals the 50% zone length divided by two and multiplied by 1.48. So, the German data for 50% zones is all one needs to define the exact shape of the normal distribution curve (bell shaped), cause the mid-point is in the middle. The different size of the 50% zones sets the shape.

The average number of rounds fired at a T34 M43 to score a hit in the math simulation was: 1 round each for 2 tanks hit 2 rounds for 1 tank hit 3 rounds each for 9 tanks hit 4 rounds each for 6 tanks hit 55 rounds fired for 18 kills Average is 3.06 rounds fired per tank kill (every round which hit had a destructive effect), not very much different from the German report you quoted (2-3 which averages about 2.5). I don't think the math model results are very much different from the German battle report with 75L43. One difference could be a higher percentage of guns with better than average dispersion, and better range estimates to start with.

In the German report it took 2-3 shots per tank, German crew firing tests required a hit by the fourth round at any range from 1200m to 2000m or so. The Germans in the report were obviously better than my math model, so maybe they had the range measured beforehand. Or maybe they had someone with a TF 14 giving them range measurements. O they were better at range estimates than my model estimated (if they were always within 200m of actual with the estimate alot more hits would occur on the second and third try. There is a reason why few hits occur on the second shot with bracketing, a fact that comes out when one goes through the math. Second shot corrections typically add or subtract 200m, which usually leads to a miss. The first shot at a 1500m target will be short or long by several hundred meters, usually more than 200m. So adding 200m or subtracting 200m from the first range estimate doesn't come close enough to score a second round hit. I did the math, looked at the results and know why the misses occurred. Only on the third shot was the range estimate very close to actual, usually after a long on the first shot and a short on the second, or vice versa. Remember that we're looking at bracketing fire, not burst on target, so the second shot is not necessarily the best unless the first round range estimate is very close to actual (which it hardly ever is).

Here are the estimated hit % for 75L48 and 88L56 APCBC against the 2' x 5' Panther turret when the range is known, and a comparison to 6 pdr results: 500 yards 89% for 75L48 96% for 88L56 89% for best 3 6 pdr 52% for worst 2 pdr 800 yards 69% for 75L48 83% for 88L56 84% for best 3 6 pdr 52% for worst 2 pdr 1000 yards 57% for 75L48 72% for 88L56 81% for best 3 6 pdr 34% for worst 2 6 pdr 1500 yards 30% for 75L48 54% for 88L56 62% for best 3 6 pdr 12% for worst 2 6 pdr Average for 88L56 is close to best 3 6 pdr guns when range is known.

Mr. Tittles, I looked into your concerns regarding the impact of the 75L48 APCBC dispersion upon shot corrections at 1500m, and here is what I found: 1. The Germans specified that bracketing be used beyond 1200m 2. Using my hand calculator I calculated a series of first round range estimates using an average error of 16% and a bell shaped curve, and then applied random dispersion effects using one times the German table figures. 20 different series were run using bracketing after first shot misses: A. 2 of 20 first round shots landed on a T34 M43 front aspect B. 1 of 18 second shots hit the target (didn't fire at previous hits) C. 9 of 17 third shots hit the target D. 6 of 8 fourth attempt shots landed on the T34 So, after four shots with bracketing 18 of 20 targets were hit. The German firing tests that are specified in the Panzertaktik book for later war crews requires that a target between 1200m and 2000m be hit with one of the first four shots when the initial range is not known. If double dispersion were used, only 9 hits would be scored within the first four shots at each range. My guess (speculation) is that the Germans would have made sure that the testing gun had close to the average dispersion, or was slightly better than average, for everyone who shot. So, it appears that German gunners on the firing range were shooting with one times the table dispersion and not two times. British firing tests and calculations for hit % against a known range target, and targets where the range estimates are in error, use one times the dispersion. Which brings into question the use of double dispersion. Here is what I think: The British firing tests with 6 pdr Churchill IV's showed that some tanks have lousy guns. At 1000 yards the best three averaged 81% hits against a known range Panther turret, the worst two averaged 34% and all five averaged 62%. If the British results are applied to the German dispersion figures one could speculate that: A. the German figures represent the average of all guns B. the average scores for the German worst guns have double (or more) times the average dispersion C. the best guns have considerably less dispersion than the average, maybe as low as half the table figures The above theory could explain, in part, why some Tigers did most of the killing. It is also possible that because of the limited edition status of Tigers compared to PzKpfw IVH, StuG IIIG and Panther, more care was given to Tiger weapon production and the variations from the average might have been less. Would German guns have been as variable as British? Perhaps not. When German dispersion figures are compared to British 17 pdr average, the German figures look reasonable. For a 1500m target and a 16% average error, the individual range estimates for first shot attempts break down as follows: 1501-1600m, 2 1601-1700m, 1 1701-1800m, 2 1801-1900m, 1 1901-2000m, 2 2001-2100m, 1 2101-2200m, 1 1401-1500m, 2 1301-1400m, 4 1201-1300m, 1 1101-1200m, 1 1001-1100m, 1 0901-1000m, 1 20 total

The British did not use a doubled dispersion when they calculated hit probabilities, and their firing trials suggest that the single dispersion was a satisfactory figure when it comes to matching firing tests with calculations. The British found that random dispersions follow a bell shaped normal distribution curve, so using that sort of curve seems to be reasonable. Whether firing tests on a nice quiet proving grounds, or shots against a simulated Tiger model on a quiet piece of Italian terrain, would simulate one's ability to aim precisely and consistently on a battle field is another matter. The other point to keep in mind is the Churchill IV trials, where three tanks out of five shot relatively straight and true, and two tanks had guns that fired very poorly and had extreme dispersions with a known range. The German dispersion curves probably represent the average of the good, the bad and the middle of the road types. Where the average dispersion is built up by the very poor guns. Double dispersion might be good for initial shots by worried or hurried gunners using less than excellent weapons.

Using the equation presented in the rangefinder article noted in the previous post this thread, the German TF 14 stereoscopic rangefinder would be capable of the following accuracy (magnification of 14, base length of 0.9m): 1000m: 4m (12m for tripled error may be practical error) 1500m: 9m (27m practical error) 2000m: 16m (48m practical error) 2500m: 24m (72m practical error) 3000m: 35m (105m practical error) The Spielberger book on Panther tanks suggests that the practical error would be three times the calculated figure, which appears to consider human errors and day to day variations in ability. If a Nashorn were firing at a target 2000m and the range estimate were based on a stereoscopic rangefinder, the range estimate would be 2048m and the first round would, on average, be 1.61m above the target bottom for a hit if all of the target were exposed (German practice was to aim at the bottom of the perceived target height and add range to bring the shot up on the target). At 3000m, the initial range estimate would be 3105m and the first shot would be, on average, 3.45m above the target bottom for a miss. However, T34's and Shermans would be about 2m to 2.2m high (hull bottom to turret top) so a minor correction would be needed for a second shot hit (shot would overfly target turret by less than 1.5m). For instance, if the Nashorn crew estimated the second shot range as 3000m (down 100m), the round would pass by the target 1.11m above target bottom for a hit (using the average trajectory). Due to round to round random scatter, individual shots would vary from the average trajectory. It appears that Nashorn were equipped with the TF 14 scissor scope, although use of them might require that the device be folded so the arms were horizontal and then held above the vehicle armor. While the TF 14 might be capable of 105m practical errors at 3000m, a trained commander using his eyes would have an average error of above plus or minus 20% for 600m. A gunner using the triangles on the gun sight, which denote 2 or 4 mil heights or widths, would see a 3m wide T34 as 1 mil width at 3000m, or a 2m high T34 as 0.67 mils. The problem with the triangle range estimation is that a 3m wide T34 might be at a 10 degree angle to the viewer, which would make the width appear to be 4m due to viewing of the front and side armor. With 5x magnification, the Nashorn gunner sight would make the T34 seem like it was 600m away to the unaided eye, too far to separate front and side armor widths for the image. The TF 14 magnification factor of 14 would make the T34 at 3000m seem to be 214m away if viewed with the eyes alone, which might make estimates of frontal width alone possible and eliminate the addition of side armor. If the Nashorn gunner sees a 4m wide T34 at 3000m, the perceived width on the gun sight would be 1.3 mils, suggesting a range of about 2300m for a standard frontal width of 3m. The triangle method would not be close. If the gunner also used the perceived height of the T34, seeing a 2m height for a T34 M43 at 3000m as 0.67 mils, the range estimate could be close to 3000m if the entire vehicle were in view and the lower area was not blocked by ground curvature or vegetation.

German experience showed that 75L48 APCBC would not penetrate the 47 degree glacis on Shermans at 1000m. This comes from panzer crews and their commanders. CMAK does appear to give high side penetration stats for German guns. My figures, which are derived from German firing tests at 30 degrees and appropriate slope effects yield: 75L48 APCBC fired at 750 m/s 135mm at 100m 130mm at 250m 123mm at 500m 116mm at 750m 109mm at 1000m 103mm at 1250m 97mm at 1500m 75L70 APCBC fired at 935 m/s 185mm at 100m 168mm at 500m 149mm at 1000m 132mm at 1500m The Easy Eight glacis (2.5" at 47 degrees from vertical) would resist German 75mm APCBC as if it were 114mm thick and vertical, which is enough to stop about half the 75L48 APCBC hits at 750m.

In actual combat, results would be even worse. The American M62 Chevrolet 76mm APCBC round had an unusually fast decrease in hardness from nose to main body, and would bulge out on certain hits. This bulging out could lead to failures when the penetration exceeded the armor resistance, making paper penetrations real life failures. So even when the penetration stats show more piercing power than the 100m mantlet on Panther, the round will often fail at a given angle and velocity. 76mm HVAP? Don't count on it too much. U.S. experience on the battlefield resulted in a conclusion that the round might penetrate the Panther glacis plate at 100 yards, but forget it at further ranges. And the 100 yard penetration would require a bad steel glacis plate with brittle characteristics. 76mm HVAP will penetrate the 60mm or 50mm lower hull front armor without too much problem. And will ice the mantlet on most hits. U.S. tests at Isigny during August 1944 showed that all American ammo for the 76mm gun, and 17 pdr APCBC, were next to useless against the Panther glacis plate at ALL ranges. 17 pdr APDS got in a few penetrations and should have been effective beyond 400 or 500 yards but the rounds came out of the barrel and yawed terribly, reducing their accuracy and penetration. 17 pdr APCBC in British tests failed to penetrate the Panther glacis AT ALL RANGES!. If 17 pdr APCBC rounds landed in a previous hit gouge, or an earlier round cracked the Panther glacis and a follow-up hit landed on or near the crack, the 17 pdr APCBC would penetrate. Hits on the machine gun ball mount would also penetrate the Panther glacis, as noted after some of the initial combats against Panthers with M10's.

Mr. Tittles, I'am beginning to believe that the doubled dispersion is too critical for guns like the 75L58 using APCBC. Makes corrections too impossible, as you point out. Maybe something a little smaller, like 1.4x for combat conditions. Waddayouthink?

On German ships the stereoscopic readers were assigned psychologists and trainers to keep their eyes and brains in best working order. Although the accuracy of the stereoscopic range finder was great during naval combat at many miles distance getting the best one could was a necessity. A 3m long stereoscopic rangefinder was capable of fantastic accuracy. If the magnification was 25x, the accuracy at 2000m would be plus or minus 2.6m! The following site has an article on stereoscopic rangefinder history, use and accuracy: http://www.warships1.com/index_tech/tech-078.htm Of interest to WW II tank fans is the write-up on "blue coating" which made the image brighter and clearer. Sounds similar to the U.S. reports that captured German gun sights were better in reduced light than American and gave better target outline.

Some Russian 76.2mm field gun units which were fighting tanks used a method they developed at Kursk to cut down on gun crew casualties. Only one crew member was allowed near the gun at a time. After the gun fired the gunner would dive into a trench and the loader would run up and do his thing. Then the loader would run for cover and the gunner would take his position and shoot. Getting and keeping good gun crews was a high priority, and they could not afford to lose several men whenever a gun was targeted and HE landed on or near the gun. While rate of fire would be less than optimal, keeping crew members alive was a higher priority.

They didn't call it first round hit percentage because it would be for any shot where the range was precisely known, which almost never occurs on the battlefield. In the next few days I'll look into how dispersion influences second and follow-up round corrections after a miss. I looked in the Sherman gunnery manual for sight adjusting of the gun, which included a procedure with a distant aiming point. To adjust for a shot at 1000 yards, it did not involve any actual shots. One checked to see that the gun and sight were laid on precisely the same point. The next section addressed ADJUSTMENT OF SIGHTS USING TESTING TARGET, which was used when distant aiming points were not available and was a second choice to distant aiming. "Place the testing target from 80 to 120 feet in front of the tank, at approximately the same height from the ground as the gun." One then "adjust the sight to put the zero range line and the vertical lind of the sight on the mark for the sight." The next procedure is ADJUSTMENT OF SIGHTS FOR 1,000 INCH FIRING. With 1,000 inch firing trials (about 83 feet or 25.4m) dispersion would not be an issue. The Americans checked the sights without having to use 1000 yard shots, and thus avoided having to deal with random dispersion in the vertical and lateral directions.

That's not what it means. If one looks at the distribution of vertical dispersions within the curve, 50% are greater than 1.0m and 50% are below 1.0m. 5% are greater than 2.82m and 95% are below 2.82m. Take a few days and think about it.

British report WO 291/762 discusses a firing test with five 6 pdr armed Churchill IV's against a target representing a Panther turret (2' high and 5' wide). The difference in accuracy between the best three and worst two is eye opening: AVERAGE HIT PERCENTAGE WITH KNOWN RANGE FOR APCBC ROUNDS Churchill.500.800.1000.1500 yards BEST 3......89%.84%..81%..62% WORST 2...52%.52%..34%..12% AVG FOR 5.74%.73%..62%..42% --------------------------------------- 75L48........90%.71%..57%..30% 75L48 hit % based on single dispersion. The German dispersion figures probably ran in a similar fashion to the British, although the small sample of five Churchills may overstate the typical differences one might observe. At any rate, it is likely that ALL tanks and guns from WW II followed a bell shaped distribution curve, with extremely accurate guns that could place shots in a close ring, and very bad guns that put shots "all over the place".

Firing on the move has a low accuracy due, in part, to the effect of random angle changes to the vehicle as it moves. If one is shooting at a target at 600m and the gun is aimed right at the center of the enemy tank, and just before one pulls the trigger the vertical angle of the tank changes by plus 1 degree due to a tiny ant hill, the shot overflies the aim point by about 10.5m! From what I've read, the Sherman gyrostabilizer did not instantly correct for vehicle angle changes as it moved but took a small time. Here is a tidbit from the Sherman gunnery manual FM 17-12 re: Firing While Moving: "Drive at a constant speed: acceleration and deceleration upset the action of the stabilizer. Drive in a straight line, otherwise the gun yaws as the tank turns. The driver warns the gunner when rough terrain is ahead. When going over rough terrain, do not "fight" the gun (attempting to keep it on target by spinning the elevating handwheel), but wait until a constant speed is regained and the action of the stabilizer has smoothed out." And how bout this: "Even with a stabilizer, the gun does not hold constantly on the target. Watch the swing of the gun through the target and fire as the proper sight setting crosses the target." So as one is charging around a bend and beating a path towards the side of that unsuspecting Tiger the stabilized gun is bouncing around even when the going is straight and the ground seems flat. Cause the tank is changing angle ever so slightly and the stabilized gun is moving around too to compensate in a never ending back and forth.

First off, they did not fire at a 2m by 2.5m target, which is a point I have made several times. The actual target was much bigger. It would help if you would stop thinking about the area within that box. Secondly, you have not looked at how the rounds would be distributed through the entire area with a 1m vertical and 0.9m lateral 50% zone. Those 50% zones (1.0m/0.9m) are associated with 68.26% zones of 1.44m/1.30m vertical/lateral. 50.00% are outside 1.00m/0.90m. 31.74% are outside 1.44m/1.30m. 25.00% are outside 1.66m/1.49m 20.00% are outside 1.85m/1.67m 15.00% are outside 2.07m/1.87m 10.00% are outside 2.37m/2.14m 05.00% are outside 2.82m/2.55m 01.00% are outside 3.72m/3.36m 00.50% are outside 4.03m/3.64m 2.07m vertical and 1.87m lateral captures 85% of the random scatter, and the size has to been approximately doubled to capture another 14.5% (4.03m/3.64m). My point has been made.

Here are some additional dispersion figures for the 50% zone based on German wartime tests (muzzle velocity in brackets): German 50mm Pak 38 HE (550 m/s) 1000m: 0.4m vert and 0.3m lat 1500m: 0.8m vert and lat German 76.2mm Pak 36 HE (550 m/s) 1000m: 0.7m vert and 0.6m lat 1500m: 1.5m vert and 1.1m lat 2000m: 2.2m vert and 1.5m lat German 88L56 HE (810 m/s) 2000m: 2m vert German 88L56 HEAT (600 m/s) 1000m: 0.7m vert and 0.4m lat 1500m: 1.2m vert and 0.6m lat 2000m: 1.8m vert and 0.8m lat German 76.2mm Pak 36 HEAT (450 m/s) 1000m: 1m vert and lat 1500m: 2m vert and 1m lat 2000m: 3m vert and 1m lat German 75mm JG 37 u.42 (Patr 38 H1/A)(355 m/s) HEAT 1000m: 0.84m vert and 0.70m lat 1500m: 2.13m vert and 1.15m lat Russian 122mm HE (800 m/s) 1050m: 0.65m vert and 0.3m lat 2050m: 1.40m vert and 0.7m lat Russian 76.2mm field gun BR354A APBC (662 m/s) 1050m: 0.55m vert and 0.55m lat 2050m: 1.30m vert and 1.20m lat

A German speaking friend of mine at work, whose father flew German aircraft at a production center during the war, translated the ballistic table column headings for me. Here are some of the more interesting ones: a. Hit Percentage for a target area of 2.5m x 2m two percentages are given, second one is in brackets The footnote associated with hit percentage in brackets translates as: " Parenthetical values valid for battle condition dispersion (double 50% dispersion)" b. Area for a target height of 2m This column gives the range where the trajectory height is 2m or less when the gun and the aim point are at the same elevation, which roughly corresponds to a battlesight type aim (aim at target bottom). The double dispersion is nothing more than a guess as to how well the aim could be placed on the target during hectic battle conditions, and doubling the dispersion has an accuracy impact equal to adding a random dispersion of 1.73 times the base dispersion. If the base scatter 50% zone is 1m, and a 50% zone dispersion of 1.73m is added, the statistical sum is 2m. If the 68.26% zone is 1.48m (1m x 1.48) and another 68.26% zone of 2.56m (1.73 x 1.48) is added, the statistical sum is 2.96m (2m x 1.48). In other words, if a bell shaped curve with a 50% zone of 1.0m is combined with another bell shaped curve with the same 50% zone, the 50% zone that results from the two curves has a length of 1.41m. To obtain a 50% zone length of 2.0m one must combine 50% zones of 1.0m and 1.73m. Assuming that battle introduces a random scatter zone equal to 1.73 times the base value is rather severe, and the doubling may be a rough estimate without much real validity. We have not been able to find anything more on the justification for double dispersion. [ August 24, 2004, 04:55 PM: Message edited by: rexford ]

"For a total single dispersion distance of 1.0m (lateral and vertically), the spread of shots is a smooth curve with 50% within 0.5m of the center. And 68% within 0.75m of the center. And 95% within 1.0m." Oops. If 50% vertical zone is 1m long (half the shots land within a 1m vertical distance, half above mid-point, half below): 50% will be within 0.50m of center of box 68.26% within 0.74m of center (used 1.48 multiplier) 80% within 0.925m of center 95% within 1.45m of center 95.5% within 1.48m of center

If a Tiger is assumed to have a frontal area of about 3.2m width and 2.6m height, taking into account ground clearance area and a turret that is narrower than the hull and tracks, the 1500m single vertical dispersion hit %'s would be (calculated from German 50% zone data an disregarding lateral line scatter and range estimation errors): The 50L60 APC would score 100%. The 88L56 APCBC would score 100%. The 88L71 APCBC would score 99%. The 75L48 APCBC would score 92%. So the British 6 pdr gun and ammo would be slightly higher than the German 75L48 APCBC, and the 17 pdr would be superior. I suspect that the British were shooting uncapped AP but who knows for sure. Against a hulldown Tiger at 1500m (0.6m turret height), the 75L48 APCBC hit % would be 31% based on the 50% zone height of 1m. This is higher than the Brits got for the 6 and 17 pdr guns and ammo, which suggests that the British results were more spread out than the German (German results bunched around middle like bell shaped curve, British results may not be as close to bell shape curve due to limited number of rounds). [ August 23, 2004, 05:30 PM: Message edited by: rexford ]

The general approach to calculating hit % against rectangle targets follows this long drawn out process: 1. take the 50% dimensions and convert to a 68.26% zone, which is one standard deviation. Say 50% zones of 1m vertical and 0.6m lateral, which become 68.26% zones of 1.5m and 0.9m. 2. Compute vertical and lateral hit probability. Assume 2.0m vertical height. Divide 2m by 68.26% vertical zone to obtain 1.33 standard deviations. Assume 2.5m lateral width. Divide 2.5m by 68.26% lateral zone to obtain 2.78. Using a standard deviation vs area table for normal distributions: 1.33 standard deviations relates to a probability of 82% that the vertical scatter will fall within the top and bottom heights of the target. 2.78 standard deviations result in a 99.5% chance that the lateral scatter will end up with the left and right limits of the target width. 3. find overall hit % Multiply the vertical probability by the lateral probability to obtain the overall hit %, or 82% x 99.5% = 81.6%. There are many ways to do the calculations but the above method is "easiest" for rectangles. It took me several months to get this straight about 20 years ago. So a 1m vertical 50% zone has an 82% chance of landing within the height limits on a 2m high target if we disregard lateral scatter. If we double the test dispersion the 50% vertical zone becomes 2m, the 68.26% vertical zone becomes 3m and the vertical hit % decreases to 50% (2m/3m = 0.667 standard deviations). In this case, doubling the vertical 50% zone decreases the vertical hit probability from 82% to 50%. And doubling the lateral 50% zone from 0.6m to 1.2m lowers the lateral hit % from 99.5% to 84% (2.5m/1.8m standard deviations). So doubling the lateral and vertical 50% zones lowers the overall hit % from 82% to 42%. Note that MANY writers indicate that the double dispersion hit percentage is what a gunner could do under a whole lot of different situations. I don't know if the double dispersion is anything but a guess as to how wide rounds with constant aim might scatter during the stress of combat.