Bloodiest day on the Eastern Front

[Art]

What was the day of the most bitter and bloodiest fighting on the Eastern Front in WWII is not merely a subject of idle interest. The answer and the stats on daily casualties in general can enhance out understanding of military operations. Unfortunately, it seems that there have been few hard data on daily losses. Below is my analysis of Soviet losses based on the Russian database of military casualties (https://obd-memorial.ru/html/). From this database I've chosen to sets of data: 1) records of killed from the casualties lists relating to all ranks and 2) records of killed officers registered by orders of the Red Army's Main Cadres Administrations. As explained in another topic due to duplicating records the database frequently finds more than 1 record corresponding to one person. For this reason the number of records found in the database can overstate actual casualties by a considerable %. However, I proceed from assumption that this overstatement factor is more or less uniform and doesn't affect daily peaks and lows. Another source of error is the fact that sometimes an exact date of casualty is absent. Instead there is a time bracket, say 10-20 June 1942. In that case the search engine sees this records as belonging to records of losses that happened both on 10 and 20 June. This mistake cannot be easily filtered out, I have no choice but to tolerate it.
A survey of monthly numbers from the dataset 1) revealed that the largest numbers of killed in action found in the database corresponded to July and August 1943:
viewtopic.php?p=2425431#p2425431
Hence it was reasonable to start from the summer of 1943. Below is the graph of daily records of personnel killed in action (data set 1) and officers killed in action (data set 2). A noteworthy thing is that both data sets provide a similar pattern of daily losses and positions of peaks mostly coincide, which attests validity of the method. Then the peaks of casualties usually coincide with that start of large operations, such as German operation "Citadel" or various Soviet offensives. It is interesting that Soviet daily losses during the defensive phase (Citadel) were smaller than during the offensive phase of the campaign. Naturally, that only valid for men killed in action. Finally, both sets of data point unambiguously to 17 July 1943 as a peak of Soviet losses in the summer campaign of 1943. We have 15744 records of killed in action (all ranks) and 1599 records of officers killed in action. That peak can be attributed to the start of the offensive at the Donets and Mius Rivers and continuing fighting in the Orel salient.
All these data describe Soviet losses only. It appears from the book of Zetterling and Francson (Kursk 1943: a Statistical Analysis) that the highest German losses in the campaign occurred on 5 July 1943 (first day of "Citadel"), but I can't be 100% sure.

[otlichnik]

Fascinating project Art. Thanks for sharing.

The fact that officer losses trend so closely with all personnel losses is very interesting, as is your observation that that peaks occur at the start of offensives versus other times.

I wonder if the "peaks at start of offensive" is a universal trend across 20th century large scale warfare or not. One could easily assume it is, but I wonder what stats or studies are out there for other armies/conflicts.

A couple of questions.

What proportion of the loss statistics fall into the time bracket category instead of the exact date category? Is it a common phenomenon?

Have you done any more case studies like your Sergey Sergeyev, and if so do they yield figures like the 70%?

Shawn

[Art]

Have you done any more case studies like your Sergey Sergeyev, and if so do they yield figures like the 70%?

The database is being updated and new data are uploaded. At the present time it finds 542 records corresponding to "Sergey Sergeyev" in casualty reports compared with 509 the last year. I can't say if earlier results are fully applicable, apparently an update of the sample study is needed.

What proportion of the loss statistics fall into the time bracket category instead of the exact date category? Is it a common phenomenon?

I didn't count, but it's nonnegligible. In any case that's the reason why the sum by days doesn't coincide with totals for month, or sum by months doesn't coincide with totals by year. It should be stessed again that the graph above gives the trend but not exact numbers. Yet I'm sure that the trend itself is rather accurate.

[otlichnik]

Have you done any more case studies like your Sergey Sergeyev, and if so do they yield figures like the 70%?

The database is being updated and new data are uploaded. At the present time it finds 542 records corresponding to "Sergey Sergeyev" in casualty reports compared with 509 the last year. I can't say if earlier results are fully applicable, apparently an update of the sample study is needed.

I was wondering if you or anyone else had done more "names" like "Sergey Sergeyev" as test examples.

Shawn

[Art]

Don't know.

A second installment: another period of intensive fighting (January-March 1944). Curiously, here all major casualties peaks seem to be associated with the northern part of the Eastern Front. Maximal numbers (11 797 KIA all ranks, 998 KIA officers on 14-15 January 1944) were smaller than during the summer campaign of 1943.

[Art]

Third part: summer of 1944:

Attribution of some smaller spikes of casualties is in question, but it is obvious that two largest spikes are associated with the start of "Bagration" and Jassy-Kishinev operations. Peak numbers for the period are 13 565/1236 killed (all ranks/officers).

[Art]

Finally, casualties in 1945

It was difficult to attribute peaks to a particular operation, especially in February-March when several operations overlapped. Interestingly, some level of losses (about 100 killed per day) was retained even after the VE day. Some people were killed in accidents, some were lost in clashes with anti-Soviet insurgents, some records had typos in the original documents. A small spike of casualties in the last days of the war (7-8 May) is also interesting (was it fighting at Prague? Elbe?)

[Reigo2]

Minor comments regarding the obd-memorial database. I have taken out all the casualties (as much as I was able to find) for Estonia from February-September 1944. Removed duplicating and erroneous records.

-My impression was that the duplication level varied - some divisions had more duplicating records, some significantly less.
-Detected (only) one instance when probably a greater amount (possibly up to 1/3) of records were missing from the database (it was about 10 days period for one divison).
-Most units had low number of erroneous records but sometimes they could be encountered in large amounts. Often the reason seemed to be that a bunch of casualties for a specific unit were added to the database based on a source which was difficult to interprete and was then interpreted in a wrong way - which resulted obviously a wrong death date. I also remeber that there was a constant problem with missing soldiers who were added to the database based on the documents which described last contact (a letter recieved by relatives) with them. Last contact may have been in 1941-1943 but database compilers chose to indicate the death date according to the document compilation date (1944).

[otlichnik]

I haven't tracked fighting by the units that remained near Berlin for 7-10 May yet, but I believe that there was certainly still enough minor combat going on around Prague until at least May 10th to account for at least part of that small spike.

[Art]

There was some fighting between Havel and Elbe on 4-7 May (a mostly forgotten episode of the war). I believe, most casualties on 7-8 May were suffered in the Prague operation, there were little combat activity elsewhere.

[Art]

Casualties in the winter campaign 1942/43.

Unlike previous installments here casualties of officers are given as a sum of search results by words "killed" and "fallen". At the start of this period (November 1942) most officer casualties were registered in the database as "fallen", whereas by the end of March as "killed". That seems to be just an artifact of data processing.

[Art]

So in the final run we have two days of peak casualties
17 July 1943 - 15 744 results found in the database by the keyword "killed"
and 14 January 1945 - 15 577 results found by the keyword "killed".

These figures were very close to each other, given duplication and errors in the database it is impossible to say what day featured the highest actual number of killed. Yet it seems very probable that the largest number of daily bloody losses were suffered on one of these two days (17.7.43 and 14.1.45). Curiously, both days are not the first which come to mind when you think about the most intense fighting on the Eastern Front. 12-14 January 1945 featured the beginning of three large offensive operations (Vistula-Oder, East-Prussia and West-Carpathian offensives) which encompassed more than 4 million men (not counting Polish, Romanian and Czechoslovak allies). The size of the forces involved evidently explains a high level of casualties.
Another noteworthy thing is that close to 1000 men were killed even in days of lull with combat activity limited to background level.

[Vasilyev]

Third part: summer of 1944:
Jun-Aug44.png
Attribution of some smaller spikes of casualties is in question, but it is obvious that two largest spikes are associated with the start of "Bagration" and Jassy-Kishinev operations. Peak numbers for the period are 13 565/1236 killed (all ranks/officers).

The Kovel' withdrawal battles in early July cost 47th Army 1,564 KIA. In comparison, the much longer July phase of the Lublin-Brest Operation cost 1,905 KIA (Subtracted 100 KIA for the quiet period between the two battles). 11th Tank Corps lost 243 KIA and 107 missing on July 8-9 according to its war diary, as well as 79 irrecoverable tanks. Definitely a not-insubstantial contributor to the spike in losses.