Kill Ratios

To address my question above, if he "ditched" in the ocean, even due to otherwise-minor damage, is he a kill?

I'd say "Yes!" The plane is a total loss and the pilot is somewhat likely to be the same.

What if the plot bails out, his chute fails, and he is killed, but the plane itself soft-lands to be recovered later. The pilot, however, died. Is THAT a kill?

GregP said:

Question for Buffnut and all the people who feel a kill means total destruction of the aircraft or the pilot.

Consider the PTO. Assume you are over water.

I would say that any damage causing the aircraft to ditch was a confirmed kill since the aircraft was lost completely to the bottom of the ocean and was not worth recovering for parts, unless it ditched in a shallow bay, right next to a crane on a barge. Some pilots were lost, some not, but planes that were lost over the ocean seem almost certainly a valid kill, even if the damage was only a bullet through a fuel line. Once in the ocean, it CAN fly gain, but not without more effort than is required to build up a plane from scratch.

The ONLY reason I can think of you would DO that is when no other option is available, such as 75+ years later when no new production is possible. If you can recover one, you can at LEAST use it as a template for making new parts. But I wouldn't call it economically a good deal to recover a working warplane if new production is still coming off the line or available in the boneyard.

Any comments on it or thoughts to the contrary?

Click to expand...

The scenario you outline is absolutely a "kill". Not sure I've ever said anything to indicate a different interpretation. Even an aircraft shot down over land that COULD be recovered (eg after an effective forced landing) counts as a "kill". Whether or not the aircraft was subsequently returned to service is irrelevant - I've seen photos of Spitfires that were, essentially, scrap but which were returned to service after factory rebuilds.

Thanks Buffnut. Just wondered.

I've always said that if a pilot shot an enemy out of the fight, particularly on escort duty, it should be a victory since he defeated the intent to stop the bomber or stop the fighter attacked from being effective. If that isn't so, then I'd surmise the attackers would LEAVE the escorted planes to pursue the potential victim and ensure a "kill."

So, my intent was to cause the escorting fighters to STAY with the escorted bombers and get the kill. What good is it to remind pilots of duty when the glory goes to the guys who get the kills? They are young and full of testosterone. No joy ... no real reason to stick around other than a sense of duty. The ones who leave to pursue and get the kill are in target-locked mode. Make THEM stay around by letting them have their victories when they remove someone from the fight. The entire intent of keeping track of kills, at least at first, was to maintain morale in the fighter pilot ranks.

Later, it was a means to estimate the remaining enemy fighter strength. But morale was the primary reason it came into being.

In the ETO, many mismatches between a victory credit and 'no actual loss' occurred when a LW a/c was seen to crash land after hits - but was less than 60% damaged by judgment and carried on LW rolls as "returned to service" after repairs.

GregP said:

The entire intent of keeping track of kills, at least at first, was to maintain morale in the fighter pilot ranks.

Click to expand...

Not in the RAF

In the very early stages of the war some pilots did receive the sort of press attention more common to WW1. Think of men like Mould, Richey or Kain (until he killed himself in an impromptu celebratory display) and some like Voase Jeff or Orton, now largely forgotten. This was for the benefit of the civilians back home, not the RAF and it was soon more tightly regulated.

GregP said:

Later, it was a means to estimate the remaining enemy fighter strength. .

Click to expand...

That was always its purpose in the RAF. There is a reason that the debriefings were carried out by intelligence officers

It is also worth mentioning that claims, even once filtered and assessed were always treated with a great deal of caution by the men who had some inkiling of what was really going on. Keith Park was notoriously sceptical of the figures accepted by the Air Ministry, his own reports contained much lower figures, yet still too high by a factor of 2-3.

Cheers

Steve

GregP said:

Thanks Buffnut. Just wondered.

I've always said that if a pilot shot an enemy out of the fight, particularly on escort duty, it should be a victory since he defeated the intent to stop the bomber or stop the fighter attacked from being effective. If that isn't so, then I'd surmise the attackers would LEAVE the escorted planes to pursue the potential victim and ensure a "kill."

So, my intent was to cause the escorting fighters to STAY with the escorted bombers and get the kill. What good is it to remind pilots of duty when the glory goes to the guys who get the kills? They are young and full of testosterone. No joy ... no real reason to stick around other than a sense of duty. The ones who leave to pursue and get the kill are in target-locked mode. Make THEM stay around by letting them have their victories when they remove someone from the fight. The entire intent of keeping track of kills, at least at first, was to maintain morale in the fighter pilot ranks.

Later, it was a means to estimate the remaining enemy fighter strength. But morale was the primary reason it came into being.

Click to expand...

Short term staying with the bombers has benefits. But the critical factor is the skill and training of the intercepting pilots. By permitting the escorts to roam the kill rate of this factor is greater. Short term the bombers may suffer marginally if the escorts roam. But long term the interceptors will degrade more with roaming -if not in numbers, in pilot quality. Long term, roaming also lessens the effectiveness of bomber interceptions.

I think that's what Doolittle had in mind.

I'm one of those folks that thinks way too many calories are burned by air combat enthusiasts with regard to kill counts/claims.

When I read Mike Spick's take on the matter I figured it was something I could get behind:

The vexed issue of overclaiming has bedeviled the question of fighter pilot scores since air combat began. In theory it should be simple to use the modern expression, the body count should give exact answers. But it doesn't. Rarely does the number of bodies match the number of claims. There are many reasons for this, most of which arise from the confusion of a fast-moving three-dimensional combat. Possibly three or four fighters attached a stricken aircraft simultaneously, or in quick succession. The pilot of a damaged aircraft might 'play dead'. leading to a mistaken claim. If the action took place over water, there might be no bodies to count. And even the most careful reconstruction often failed to reveal who did what with any degree of certainty.

If we are to avoid a nonsense on the subject of scores, we must avoid the expression 'kill', and substitute 'victory', defining a victory as an event which occurs when an enemy aircraft is defeated in combat in circumstances where the victor believes that it will be a total loss. Only then are we on surer ground.

Hi Bill,

I've always disagreed with that one. If you shoot someone out of a fight, it should be a victory. I really do not understand the mentality that the aircraft must be destroyed. It is the job of the escort to protect the escorted, and shooting someone out of a fight seems to fulfill that job, at least to me.

Of course, maybe that's why the fight over victories has lasted so long, huh? I still disagree with the Boyington debate. In many files, Lance Wade is credited with 23 victories, making him the 13th best U.S. Ace. But all his victories were in RAF service. If his victories in the service of another nation count, so do Boyington's with the AVG. I don't want to fight that fight in here. Let's just say that resolution doesn't seem much closer than it ever did.

I've always tried to look at victories from the perspective of what would pass statistical muster. That is, anything that can be analyzed should be a statistically significant event and randomly-selected. That is why single missions are irrelevant. They CAN be chosen from the population, to be sure. But choosing them because they were good is classic statistical bias, and results in incorrect conclusions. That's also why I discount the Finn experience with the Buffalo. It was such a small, stratified sample of Buffalo operations as to be irrelevant in the face of other Buffalo experience. It doesn't take away the skill of the Finns at all. It means their experience was atypical of other experiences with the aircraft. Their skill is undeniable.

Sometimes, rare events happen. Doesn't mean they are statistically significant or that they should affect the general perception of what to expect in future events. Hitler could have had a heart attack early in the war. But counting on it because your neighbor had one would be the height of foolishness.

Hope your analysis of the MTO is coming along well. All the best.

Greg,

The challenge with the statistical analysis you're proposing is that it would only ever work for aircraft that saw air-to-air combat in very large numbers. ANY airframe that was used in smaller numbers is likely to have the results heavily skewed by a few missions/events...and that spans everything from the Blackburn Skua in 1939 to the Mitsubishi J2M in 1945. There simply isn't a sufficiently large sample for these types from which to draw any objective statistical conclusions.

Selectively discounting one set of data just because you don't like it runs the risk of even worse subjective analysis than we often see in such debates. You mention discarding Finnish use of the Buffalo because the results were atypical but why not discount the US employment of the type because it only saw one fighter-on-fighter operational engagement on one day, at Midway? From a statistical perspective, that's vanishingly trivial given the scope of combat for US forces in the period 1941-1945.

When dealing with limited-use aircraft types, you have to find some other method of describing their effectiveness. Statistics won't cut it so we, unfortunately, must look across the available evidence and examine it all, evaluating (as far as is humanly possible) why things happened the way they did. That's where the fun lies. I'm afraid kill ratios and totals of claims made by individual pilots are of zero interest to me because we can't really learn anything from them, except in the very broadest sense. They can be a starting point for the "why did it happen like that?" question, but that's about the limit of it.

Just my two penn'orth.

Cheers,
Mark

Greg - after a long delay, a new book project is forcing me to deal with the MTO.

As to Victory credits, the eternal discussion relies on at least factors to close on statistically relevant results. Pilot recognition of the inability of the claimed aircraft to continue to (including pilot bail out) or recover from what seems (to him) a fatally damaged aircraft. The tactic of FW 190 pilots to intentionally spin when in tough situation is an example. Valid confirmation by an equally observant and objective second party or confirmation of the damage observed as 'fatal' by an objective third party reviewing adequate quality combat film. Pilot integrity.

The loss of a wing or empennage or engulfed in flames or exploding in mid air or crashing were criteria for AAF credits. That said, there is no question that IO's were under pressure to grant credits when some doubt as to actual status of the target still remained. The category of Probable was the 'frequent out'.

Ahh, Boyington. Simply, the records were lost when AVG retreated Jan-Feb 1942. Boyington was paid for the ones the AVG reviewing authority granted as destroyed. He go paid for fewer than he later claimed. Take what you want and leave the rest.

Regards,

Bill

Bill,

Pressure on the IOs is not the only factor. In a messy, churning air combat, it's very difficult for any pilot to be certain of a particular result. Watching your adversary until he crashes is a certain recipe for you, yourself, to be bounced and probably shot down by his wingman. Having 2 pilots watching the same aircraft crash simply doubles the chance that one or both will, themselves, be shot down. All too often, the pilots interpreted smoke and a rapid departure from normal flight as a sign that the aircraft and/or pilot was destroyed when, in reality, it was often just the adversary firewalling the throttle and taking evasive action.

I'm not criticizing those involved, simply reflecting the difficulty of confirming "kills" in combat situations. If it were otherwise, we wouldn't see the considerable overclaiming by all sides during WWII.These messy human factors are precisely why I have difficulties using any broad statistics on "kills" or kill ratios.

Cheers,
Mark

When judging the accuracy of "claims" it is well to remember that something like over a dozen He 113s were claimed to have been shot down during the summer/fall of 1940 over England and the Channel.

Hi Buffnut,

Ther number of aircraft participating is irrelevant. If only 100 fighters participated in air combat, then a good sample is obviously any number greater than, say 39 or so. Doesn't mean thousands; means a significant chunk of the population available. Statistics don't require a set number. They require a set fraction of the population in question.

The biggest stumbling block is lack of available data.

The U.S.A is the ONLY country who did a study of aerial victory claims after the war. So, we are left with wartime claims for the rest. My contention is that if we must use wartime claim data for everyone else. we should use wartime claim data for the U.S.A., too. Apples to apples. I have a VERY GOOD file of German claims, opinions to the contrary notwithstanding. It isn't 100% complete because some data have been lost. But it is VERY good, considering the data available and covers some 67,000+ claims, making it at LEAST a very good statistical sample of claims.

The real problem is getting good claim data from everyone. Soviet data are scarce. Japanese data are almost non-existent.

The Magnus Family website (Ace file attached - sorry for the duplicates. Maybe a mod can delete it down to one file. It has Magnus claims and my own analysis of USAAF Report 85 at the beginning) is good for more or less official victory claims, and Jan Safarik's website is good, though lacking in many areas. Neither one 100% agrees with the other. People tell me. "it's simple, just use these 8 volumes to get Pacific claims for some time period." I am not into buying a library to get good claim data.

Bill Marshall uses MACRs and I am very sketchy about the way to get them and what they actually say.

I am willing to swap files with anyone who is interested in dealing with this and/or working with someone to analyze the data when we get a "good data set." (I do have a 20Mb file size limit in email) My intent is NOT to be 100% accurate, but rather to have a good picture of what went on. Many claim files tell us the victim that was shot down, but not what plane the victor was flying. The U.S. Navy is scrupulous in NOT saving the same data as the U.S.A.A.F., maybe intentionally because they don't want to be compared together, who knows?

It is frustrating and may never be resolved to the satisfaction of everyone. I am wanting to get Bill's new volume when he publishes it and I'd take HIS data for the U.S.A. any day. Now if we can just get him to analyze the UK, Japan, Germany, and the Soviet Union, I'd have what I was looking for!

Best regards, -Greg

GregP said:

Ther number of aircraft participating is irrelevant. If only 100 fighters participated in air combat, then a good sample is obviously any number greater than, say 39 or so. Doesn't mean thousands; means a significant chunk of the population available. Statistics don't require a set number. They require a set fraction of the population in question.

Click to expand...

But the smaller the population, the greater the risk that a non-representative fraction may skew the analysis in either direction. You indicated a desire to remove the Finnish experience with the Buffalo which would remove 44 airframes from a (rough) total of 320 that saw actual combat. How is removing such a large proportion a valid basis for any form of realistic analysis?

I think the Finnish Buffaloes show the limitations of a kill ratio. The same aircraft type with different pilots in a different tactical situation achieved a totally different result.

Also is a Spitfire I the same aircraft as a Spitfire 22? They're not even close in performance. What about a Mustang I in British service compared to a P51D in American service? Are they the same aircraft? Different engines, different performance, even used in totally different roles. Do you put their kills in the same ratio?

Greg - Fold3.com has 99% of the ETO MACRs. Thank you for the kind words, but even MACR based research leaves room for error simply because even the 'witness accounts' speculate on cause of loss. When I compliled 8th AF losses in my book I separated air to air as 'confirmed' or 'higher probability' when the witness last saw one of his squadron mates being chased by enemy fighters but didn't see what actually happened. Ditto 'flak' loss when for example a Mustang crashes into aircraft being strafed or hits trees during the run. Could have been loss of control while trying to avoid flak.

There are many 'non-absolutes' in the assessments.

Thanks Bill!

Buffnut,

I do NOT wish to remove the Finnish experience. I want ALL the experience lumped together and then we take a random sample. If the Finnish experience is partially selected, fine. What I don't want is a sample of really good missions selected BECAUSE they were good experiences. That is statistical bias. Tne sample selections must be done not knowing what data they contain at all, and must be chosen randomly. If not, then statistics cannot tell you anything of value.

ALL conclusions from statistical process depend upon random samples from populations where each and every member has an equal chance of being selected. If not, the math is invalid.

An obvious example of that are political polls taken at a shopping mall. The actual population is only those people who shop that mall; not the county, state, or country population.

Sorry, it's the engineer in me; and I shall desist.

ricardo said:

Hi everybody.

I'd wish to know if someone has info about kill ratios (air to air victories). I understand that Hellcats has an astonishing 19:1 over the IJN. I've read that during BoB in fighter versus fighter dogfights, the Bf-109 was the undisputed champion. 333 Bf109 lost compared to the 272 Hurricanes and 219 Spitfires.

Against Spitfires: 219 to 180
Against Hurricanes: 272 to 153

Click to expand...

Well, the Hurricanes were tasked with chasing bombers, so you can't do a comparison on part of the battlefield. "My tank division wiped out all the other tanks, but we lost the war."

ricardo said:

Hi everybody.

I'd wish to know if someone has info about kill ratios (air to air victories). I understand that Hellcats has an astonishing 19:1 over the IJN. I've read that during BoB in fighter versus fighter dogfights, the Bf-109 was the undisputed champion. 333 Bf109 lost compared to the 272 Hurricanes and 219 Spitfires.

Against Spitfires: 219 to 180
Against Hurricanes: 272 to 153

Click to expand...

Battle of Britain - aircraft losses

David Fred said:

Battle of Britain - aircraft losses

Click to expand...

There are lies, damned lies and statistics.