Changes in the national composition of the Red Army in 1942-1945: multidimensional statistical analysis of data taking into account various categories of military personnel

Bezugol'nyi Aleksei Yur'evich Doctor of History Leading Researcher, Institute of General History of the Russian Academy of Sciences 119334, Russia, g. Moscow, ul. Leninskii Prospekt, 32a besu111@yandex.ru Borodkin Leonid Iosifovich Doctor of History Professor, Department of Historical Informatics, Lomonosov Moscow State University 119992, Russia, g. Moscow, ul. Lomonosovskii Prospekt, 27 korp. 4, kab. 454G lborodkin@mail.ru

Leontyeva Nadezda Postgraduate student, Historical Information Science Department, Lomonosov Moscow State University 119192, Russia, Moskva oblast', g. Moscow, ul. Lomonosovskii Prospekt,, 27, korp. 4 nadleon96_96@mail.ru Other publications by this author

Introduction In studies on the history of the Great Patriotic War, statistical materials are often used that characterize certain assessments of the state of the Red Army: its regular and list composition, weapons, equipment and many other indicators.

The study of the national composition of the Red Army based on statistical historical sources and with the help of statistical methodological tools has not been carried out until recently. Episodic appeals to national statistics on the recruitment of the Red Army during the war years, as a rule, only clouded the picture, since random, irrelevant figures from heterogeneous sources were involved in the analysis, based on which equally unreasonable, unrepresentative conclusions and judgments are made. This often becomes a favorable environment for politicized speculation and speculation about the extent of the participation of various nationalities in the wars of the Soviet Union, their contribution to the cause of victory, losses, disciplinary and award practice, military exploits and crimes of representatives of a particular ethnic group, etc.

Until recently, a significant number of statistical sources reflecting the socio-demographic characteristics of the Red Army personnel were classified. Currently, the Central Archive of the Ministry of Defense of the Russian Federation (CAMO of the Russian Federation) has consolidated albums of socio-demographic data of the Red Army personnel. As a document of particular importance, they were executed once every six months in a single copy and as a generalizing reference material were intended for the top leadership of the state and the armed forces. Complete uniformity in the content and design of albums is noted in the last period of the war – since 1944, some of the albums have been preserved in the TSAMO in the post-war edition. They were compiled according to original documents by the long-term head of the General Staff body for the statistical accounting of personnel (at various times bearing various names), Major General of the Quartermaster Service S. M. Podolsky. Currently, such consolidated documents have been identified and put into scientific circulation as of: January 1, 1941, July 1, 1942, January 1, 1943, July 1, 1943, January 1, 1944, July 1, 1944, January 1, 1945, October 1, 1945. These materials are stored in the inventories 24, 26 and 32 of the fund 7 of the CAMO. All the listed materials of the foundation 7 were declassified only in the middle of 2017 and have not yet been known to the historical community. The albums for July 1, 1941 and January 1, 1942 remain hypothetically undiscovered to date. Relative to the first date, there are indirect indications in the sources that such a document was not prepared. At least, it is certainly known that the General Staff of the Red Army did not have accurate data on the size of the Red Army until the autumn of 1941, due to the inability to obtain operational data from the troops. In addition, drawing up a report document on the state of the army, which was in the process of mobilization deployment in the initial period of the war, did not make much practical sense.

One way or another, the researcher has recently had in his hands a unique mass material that meets the basic requirements of statistical observation (periodicity, universality of coverage, qualitative uniformity of statistical units, uniformity in album compilation techniques, etc.). Historical and statistical research of this material is already underway and leads to important conclusions regarding national representation in the Red the army during the war, trends and driving forces of the change in the national composition of the Soviet troops ^[1].

An integral part of the considered summary albums for the period of the Great Patriotic War are statistical data reflecting the national characteristics of various categories of military personnel: commanding officer (since 1943 - officer), junior commanding officer (since 1943 – sergeant), private, cadet staff.

The purpose of this work is to analyze the changes in the national composition of various categories of Red Army servicemen that occurred from mid-1942 to early 1945 (this period is determined by the availability of source data). Based on the nature of the source under study, the proposed research approach is based on the use of cluster analysis – one of the most well-known methods of multidimensional statistical analysis.

About the research methodology When studying mass historical sources, the initial information can often be presented in the form of a set of objects, each of which is characterized by a number of features (indicators).

The number of objects can reach many tens, hundreds or thousands; the number of features can also be quite large and number in the tens. Obviously, direct (visual) analysis of the data matrix with a large number of objects and a set of features is practically ineffective – you can only identify individual features of the structure under study, extract illustrative, particular examples. Methods of analyzing the structure of a set of objects form a set of methods of cluster analysis (multidimensional classification). In the context of the growing attention to the possibilities of data Science, the role of cluster analysis as a tool for identifying the structure of a multidimensional typology of objects has been noticeably updated.Cluster analysis

The traditional method of constructing a typology is usually reduced to grouping the studied objects on the basis of one (maximum two or three) features.

It is important to note that the traditional methods of typological grouping are aimed at identifying qualitatively homogeneous groups of objects by defining the boundaries of intervals on the axis of one of the group-forming features; these techniques are informal and are carried out on the basis of meaningful concepts and the experience of previous studies.

The modern level of development of methods of multidimensional quantitative analysis and computer technologies allow classification on a broader and objective basis – taking into account all significant structural and typological features and the nature of the distribution of objects in a given system of features. This classification is based on the desire to collect similar objects in a certain sense into one group, and in such a way that objects from different groups are as dissimilar as possible. Such methods are called methods of multidimensional classification (cluster analysis, taxonomy).

Let's briefly consider the two most commonly used methods of cluster analysis. They are also used in this work.

        Agglomerative-hierarchical method. We will assume that all m features are measured on a quantitative scale. Then each of the n objects can be represented by a point in the m-dimensional feature space. The nature of the distribution of these points in the considered space determines the structure of similarities and differences of objects in a given system of indicators. The similarity of objects can be judged by the distance between the corresponding points. The meaningful meaning of such a concept of similarity means that the objects are closer and more similar in the aspect under consideration, the fewer the differences between the values of the same indicators.To determine the proximity of a pair of points (objects i and j) in multidimensional space, in the case of quantitative features, the Euclidean distance is used, equal to the square root of the sum of the squares of the differences in the values of the same indicators taken for this pair of objects: 

                 (i, j = 1,2,...,n),

where d _ij is the Euclidean distance between the i-th and j-th objects, x _ik is the value of the k-th feature for the i-th object.

Note that the distance between objects depends on the "scale" of features: features whose range of values (and it depends on the units of measurement) is large, play a big role in calculating the distance between objects, unlike features whose range of variation is small. For this reason, the data is usually normalized, i.e. all the signs lead to a standard form with an average value equal to zero and a standard deviation equal to one. In this case, it was not necessary.

After calculating the distance values for all pairs of objects, we get a square matrix D of size n x n (distance matrix); this matrix is obviously symmetric.

The distance matrix D serves as the basis for the implementation of the agglomerative-hierarchical method, the main idea of which is to consistently combine grouped objects – first the closest, and then more and more distant from each other. The classification procedure consists of successive steps, at each of which the two closest groups of objects (clusters) are combined.

There are various ways to determine the distances between clusters (distinguishing methods of cluster analysis). Usually, the proximity of two clusters is defined as the average value of the distance between all such pairs of objects, where one object of the pair belongs to one cluster and the other to another: where D 2 pq is a measure of proximity between the pth and qth clusters; X p is the pth cluster; X q is the qth cluster; n p, n q is the number of objects in the pth and qth clusters, respectively.

At the first step of the procedure of the agglomerative-hierarchical method of cluster analysis, the initial matrix of distances between objects is considered, and the minimum distance is determined by it; further, the closest objects located at this distance from each other are combined into one cluster, the row and column corresponding to the first of these objects are crossed out in the matrix, and the distances from the new clusters to all other clusters (at the first step – objects) are calculated according to the above formula – as the average of the distances from the objects of the first cluster to each of the others. These newly calculated values are entered in the row and column of the distance matrix corresponding to the second object from the first cluster.

At the second step of the procedure, using the distance matrix reduced by one row and one column, the minimum distance is again determined and a new cluster is formed. This cluster can be built as a result of combining either two objects or one object with the cluster built in the first step. Next, one row and one column are crossed out in the distance matrix, and one row and one column are recalculated, etc.

Thus, the hierarchical method of cluster analysis includes n–1 similar steps. At the same time, after each step is completed, the number of clusters decreases by one, and the distance matrix decreases by one row and one column. At the end of this procedure, you will get one cluster combining all n objects.

The results of such classification are often depicted as a dendrogram (a tree of hierarchical structure) containing n levels, each of which corresponds to one of the steps of the described process of sequential cluster enlargement.

     The k-means method. Another method of cluster analysis is the so-called k-means method. Unlike the agglomerative-hierarchical method, which does not require a preliminary estimate of the possible number of groups of objects, this method is based on the hypothesis of the most likely number of classes. The task of the method in this case is to build a given number of clusters that should differ from each other as much as possible.The classification procedure begins with the construction of a given number of clusters obtained by randomly grouping objects.

This is followed by an iterative process of moving objects between groups in order to minimize the total intra-class variance of indicators and maximize inter-class variance (i.e., each cluster should consist of the most "similar" objects, and the clusters themselves should be as "unlike" each other as possible).

The results of this method allow us to obtain the centers of all classes (as well as other parameters of descriptive statistics) for each of the initial features, as well as to see a graphical representation of how and by what parameters the resulting classes differ.

The analysis of statistical data on the evolution of the national composition of the Red Army during the Great Patriotic War is carried out below using both described methods of cluster analysis. Analysis of changes in the national composition of the Red Army during the war

Let's first consider which nationalities prevailed in the Red Army among officers and enlisted personnel, comparing the data for July 1942 and January 1945 (see the tables in the Appendix).

It is clear that, in general, those ethnic groups dominated whose share in the population of the USSR was the highest. However, when considering statistical data specifically, the picture turns out to be more complex, which is reflected in Table 1, which provides statistics for a number of the most characteristic ethnic groups.

     Table 1. Changes in the representation of a number of nationalities in the categories of "privates" and "officers" of the Red Army (July 1942 – January 1945) Nationalities

As can be seen from Table 1, during the two and a half years of the war there were significant changes in the national composition of officers and enlisted men of the Red Army. The share of Russians in the "ordinary" category decreased from 68.3% to 58.4%, while the share of Ukrainians in this category more than doubled – from 10.7% to 21.7%, and the share of Belarusians – three times. At the same time, the share of all three Slavic peoples in the category of "officers" has hardly changed. The proportion of Kazakhs, Uzbeks, Armenians, Georgians, and Tatars among the rank–and-file has noticeably decreased (by one and a half to three times); the exception here is Moldovans, whose share in the rank-and-file has increased from 0.06% to 2.55%. Attention is drawn to the relatively high proportion of Jews in the officer corps at both time sections (5.4% and 5.3%, respectively; according to this indicator, only Russians and Ukrainians were ahead).

The interpretation of these structural changes in the national composition of the Red Army requires a detailed analysis of a number of factors. Let's mention the main ones here. During the period of the war under review, there were major changes in the number of military personnel: in July 1942, the Red Army numbered 10.15 million people, and in January 1945 – 12.62 million; at the same time, irretrievable army losses during the war years, according to official data, amounted to 8.688 million people. ^{[1, pp. 33-34]}. Over the years, the order of conscription for conscripts of a number of nationalities has changed, several peoples have been deported, the composition of the army has been significantly supplemented by conscription from the liberated territories of the USSR, etc.

Another perspective of the analysis of changes in the national composition of the Red Army from July 1942 to January 1945 is also of interest. We are talking about changes in the structure of the distribution of representatives of each nationality into four categories of military personnel: officers, sergeants, privates, cadets. It is in this "optics" that the percentage data in both tables of the appendix to the article are presented. The scale of these changes is characterized by the data in Table 2, illustrating structural shifts in the personnel composition of the army for a number of nationalities.

Table 2. Changes in the personnel structure of the Red Army for a number of nationalities (July 1942 – January 1945).Nationalities

As follows from Table 2, by the beginning of 1945, the share of officers and sergeants among Russian servicemen had increased, and the share of enlisted personnel had significantly decreased (from 66.7% to 58.25%), while among Ukrainians and Belarusians the trend was the opposite: the share of officers decreased by about one and a half times for Ukrainians and by two Belarusians have twice, and the share of ordinary personnel for Ukrainians and Belarusians has increased significantly.

A noticeable increase in the share of officers and sergeants and a decrease in the share of enlisted personnel characterizes military personnel -Tatars, Armenians, Kazakhs, Ossetians. The opposite trend is expressed for Lithuanians and Moldovans: for example, among Lithuanians, the share of officers decreased from 16.5% to 2.2%, and the share of sergeants – from 23.3% to 6.7%, while the share of privates increased from 59.7% to 90.4%. These changes affected Moldovan servicemen even more (the share of officers and sergeants decreased from 27.6% to 2.3%, and the share of privates increased to 97.7%).

It should be noted that the share of cadets decreased during the two and a half years of the war in almost all national groups considered in Table 2.

Of particular interest is the analysis of the changes that took place during the war in the category of "officers". Let's consider the list of those nationalities for which the share of military personnel in this category exceeds 10% in each of the two time slices (Table 3). For example, we note that according to Tables P1 and P2 of the Appendix, on July 1, 1942, the number of officers of the Red Army, Russians by nationality, was equal to 860,247 people., which accounted for 12.3% of the total number of servicemen, Russians by nationality; on January 1, 1945, these figures reached 119,8532 people and 15.27%, respectively. At the same time, the share of officers in the Red Army as a whole was 12.41% in the first time slice and 13.96% in the second.

Table 3. Nationalities for which the share of officers exceeded 10% of the total number of their military personnel (in descending order).

As we can see, five ethnic groups included in this category retained their increased representation in the officers of the Red Army both in 1942 and in 1945 (these are Jews, Ossetians, Russians, Ukrainians, Belarusians). Four national groups included in this category (by the threshold of 10%) in 1942 were not included in the list of the beginning of 1945 (Lithuanians, Latvians, Poles, Moldovans). Five national groups that were not included in the 1942 list were included in this list by the beginning of 1945 (these are Armenians, Georgians, peoples of Dagestan, Azerbaijanis and Udmurts).

In general, the data in Table 3 show a fairly dynamic picture of changes in the list of nationalities with the share of officers exceeding 10%. 

Note that when compiling Table 3, the nationalities represented in the Red Army by a relatively small number of servicemen were not taken into account. As the threshold number, the value of 4,000 servicemen of a particular nationality was chosen on January 1, 1945; this is 0.03% of the total number of the Red Army on that date. The dynamics of structural changes for these nationalities can be traced according to tables P1 and P2 of the Appendix.

                          Multidimensional statistical analysis of statistical data                                                                                             about the national composition of the Red Army during the war

The detailed analysis of changes in the national composition of the Red Army during the war years can be continued further, but there is an opportunity to develop this analysis based on a historical and typological approach. The methodological basis for the implementation of such an approach can be multidimensional statistical analysis, an effective tool of which is the cluster analysis described above. The purpose of such an analysis in this case is to identify sufficiently homogeneous groups (clusters) of nationalities for each time slice, while noticeably differing in the structure of the distribution of nationalities across four categories of military personnel, i.e. systematically, and not by individual parameters. The study of changes in the composition of the identified typological clusters and their parameters on two time slices will allow us to give a generalized characteristic of the dynamics of the national composition of the Red Army.The initial data for cluster analysis are presented in tables P1 and P2 of the Appendix.

These data made it possible to form two tables (for two slices), the number of rows in which is determined by the number of selected nationalities (in accordance with the threshold value indicated above: 0.03%), and the number of columns is equal to four – by the number of categories of military personnel (officers, sergeants, privates, cadets).  Each number in the table is equal to the percentage of persons of one or another category of military personnel of the corresponding nationality.  Russian Russians by nationality, for example, in the table for 1942, the number 12.3 is in the cell "officers, Russians by nationality" – this is the percentage of officers from the total number of military personnel, Russians by nationality in the specified year (see the explanation to Table 3).

Cluster analysis was implemented using the STATISTICA package, two methods described above were used sequentially. At the first stage, hierarchical clustering was carried out, allowing to visualize the structure of the typology, to put forward a hypothesis about the number of clusters. At the second stage, clustering is carried out using the k-means method for two typological groups (clusters), while each cluster is given a statistical description. Each of the clusters constructed includes nationalities characterized by a similar distribution structure in four military categories.            

 Results of cluster analysis according to data for July 1942Figure 1 shows clustering by the hierarchical method.

The higher the level at which the unification of objects (nationalities) takes place, the less close is their distribution into four military categories. Fig. 1. Hierarchical clustering of nationalities in the Red Army (July, 1942).

Let's move on to the results of clustering using the k-means method, with the number of clusters k=2. The first cluster contains 9 nationalities, the second – 23 (Table 4). The numbers given in the column characterize the degree of "typicality" of cluster objects: the smaller the value of this indicator, the closer the object is to the conditional "center" of the cluster. Table 4. Composition of clusters by the k-means method (as of July 1, 1942) 

Cluster 1                                  

                             Cluster 2 Tatars

      As follows from Table 4, the first cluster includes Slavic peoples, as well as Moldovans, Lithuanians and Latvians, Jews and Ossetians. The most typical nationality in cluster 1 is Ukrainians, the least typical is Jews. The second cluster mainly includes the peoples of the Volga region, the Urals, the Caucasus and Central Asia. The most typical nationality of the 2nd cluster is Tatars, the least typical is Kyrgyz. Let us now turn to the statistical characteristics of the constructed clusters.

Table 5 contains the average values of the representation of the nationalities of this cluster for each of the 4 military categories (as a percentage), as well as the degree of their variation (the greater the value of the coefficient of variation in the last column, the higher the degree of heterogeneity in the corresponding category).

Table 5. Average percentage of 4 categories of Red Army servicemen for two clusters and intra-cluster variation (as of July 1, 1942)  Cluster 1

       Percent

                                                  Cluster 2Percent

 As follows from Table 5, clusters vary greatly in their characteristics. The average share of officers by nationality of the 1st cluster is 18.8%, and for the 2nd cluster – only 5.8%. For sergeants, these shares are 16.8% and 10.6%, for cadets – 3.26% and 3.06%, respectively. But in the "ordinary" category, the ratio is the opposite: in the first cluster, their share is 61.1%, in the second – 80.5%. Obviously, the nationalities of the 1st cluster occupied higher positions in the context of 4 military categories. Figure 2 contains a visualization of these differences.Note one more conclusion from the data in Table 5: the first cluster is characterized by a higher degree of variation in all 4 categories, it is especially high for the categories "officers" and "privates".

Fig. 2. Visualization of the average values of the percentage composition of 4 categories of military personnel for two clusters (July 1942)

Results of cluster analysis according to data for January 1945Figure 3 shows clustering by the hierarchical method.

It is easy to see noticeable differences in dendrograms constructed according to data on the national composition of 4 military categories in 1942 and 1945. 

Fig. 3. Hierarchical clustering of nationalities in the Red Army (January 1945).

These differences are manifested primarily in the composition of clusters (see Table 6). Table 6. Composition of clusters by the k-means method (January 1945) 

  Cluster 1              

                              Cluster 2 Ukrainians

 As follows from Table 6, the first cluster includes, as before, Russians, Ossetians and Jews, but now also Armenians, Georgians and Komi (at the same time, 6 nationalities that were part of the 1st cluster in 1942 moved to the 2nd cluster).  The most typical nationality in cluster 1 is Ossetians, the least typical is Jews. The second cluster includes the remaining 23 nationalities. The most typical nationalities of the 2nd cluster are Belarusians, the least typical are Moldovans. Let us now turn to the statistical characteristics of the constructed clusters (Table.

7 and fig. 4).

 Table 7. Average percentage of 4 categories of Red Army servicemen for two clusters (January 1945) Cluster 1

                                                 Percent

                                                    Cluster 2 Percent

 Fig. 4. Visualization of the average values of the percentage composition of 4 categories of military personnel for two clusters (January 1945)

As follows from Table 7, clusters in this case also differ markedly in their characteristics.

  The average share of officers by nationality of the 1st cluster is 21.5%, and for the 2nd cluster – only 7.6%. For sergeants, these shares differ less noticeably – 20.5% and 15.25%, for cadets – more noticeably than in 1942 (2.46% and 1.46%, respectively). In the "ordinary" category, the ratio remained approximately the same: in the first cluster, their share has now dropped to an average of 55.5%, but in the second it has also dropped to 75.74%. Summarizing, we can say that according to the data for January 1945, there was a group of nationalities in the Red Army (1st cluster) that occupied higher positions in the context of 4 military categories. Fig. 4 contains a visualization of these differences.Intracluster variability according to the data for January 1945 has its own specifics.

Table 7 shows a very high degree of variation for the categories "officers" and "privates" of the first cluster. At the same time, the variation of the "sergeants" category for the nationalities of the first cluster is much lower than for the second. The variation of the "privates" category in the second cluster increased sharply in comparison with the variation according to the 1942 data. In general, both clusters have become less homogeneous.

* * *

The analysis of statistics from the summary albums of socio-demographic data of the Red Army personnel, stored in the Central Archive of the Ministry of Defense of the Russian Federation, carried out using statistical methods, reflects the main directions of changes in the national composition of the Red Army, revealing the general and special in the dynamics of the distribution of servicemen of various nationalities by military categories.

As it is already clear from the above text, significant changes in the composition of clusters over the two and a half years of the war (from mid-1942 to early 1945) cannot be explained linearly based on a single factor or cause. A number of powerful factors influenced the circumstances of the recruitment of the Red Army troops, the first of which should be called the unprecedented scale and intensity of military operations, which required the search for appropriate human replacements for new formations and replenishing the losses of the active army. The popular propaganda stamp about the "inexhaustibility of Russia's human resources", widely disseminated by our enemy during the war years, is not confirmed by any sources. On the contrary, already at the beginning of 1942, all the military-trained, i.e. militarily trained human resources of the Soviet Union were exhausted. Untrained conscripts of all ages went into action, the number of female servicemen continuously increased, early conscription of young people was carried out. Suffice it to say that during the war years, six regular draft ages were called up (from 1922 to 1927 years of birth) instead of four. In the autumn of 1944, when the last age conscription of wartime took place, 17-year-old boys born in 1927 went to the army, while according to the current conscription law of 1939, the conscription period for most of them came only in the autumn of 1946.

All this indicates that the number of citizens fit for military service does not directly correlate with the total population of the country, no matter how big this country may seem. It is mediated by the gender, age, social, national structure of society, physical form, the level of general education and military training of the military-bound population and other factors, among which the factor of political trust in certain social and national groups on the part of the totalitarian Soviet state should be emphasized.

The ethnic structure of society as a factor that directly influenced the principles and methods of manning the Red Army, rose to full height with the beginning of the war – only general mobilization was able to "awaken" it to life. After all, the compact peacetime army from the beginning (we are talking about the tradition of the tsarist army) was Russian-Slavic in composition. The enemy's capture of the densely populated regions of the RSFSR for a short time, completely – the territories of Ukraine and Belarus – broke the usual way of manning and caused drastic changes in the ethnic portrait of the Red Army. The troops were actively replenished by natives of the Caucasus and Central Asia. After a radical change in the war, the replenishment also came from the liberated territories, again changing the national structure of the army.

The lack of human resources forced the Soviet state to change the fundamental approaches in the recruitment of the armed forces, including getting rid of the restrictions imposed by the deep indoctrination of the Red Army construction process. Therefore, if at the beginning of the war the authorities could neglect many so–called "Westerners" – residents of the newly annexed Western regions, using them in rear units or in working battalions, then in 1944-1945 they were drafted into the ranks of the Red Army at least a million people, and all "Westerners" were sent to the front as soon as possible. the front line – there was no one else to fight.

Special attention should be paid to the analysis of the ratio of privates and officers among various ethnic groups. The training and military service of command and enlisted personnel was determined by a number of fundamental and particular, temporary factors that largely coincided. For the command staff, the fundamental factors were the general level of education and involvement in the Russian cultural and linguistic environment, which gave the officer obvious advantages in professional and career growth. A striking example (except for the Russians themselves) should be called Jews, who had the highest level of education among the ethnic groups of the Soviet Union, as well as a high degree of Russification. The same, although to a lesser extent, applies to Ossetians, among whom the proportion of officers was also abnormally high. Jews and Ossetians were consistently present in the first clusters in 1942 and 1945 .

An example of the influence of temporary factors are Armenians and Georgians, whose appearance in the first cluster in January 1945 was not accidental: young people were specially recruited from representatives of these two nationalities to train middle-level commanders in numerous national formations. Although the training of national commanders was conducted in many directions, it was conducted most successfully and massively among Armenians and Georgians, which, in turn, is explained by the high proportion of people with secondary and higher education among these ethnic groups.

It is no coincidence that Ukrainians and Belarusians disappeared from the first cluster in 1945. It is also explained by a number of incoming factors of the second half of the war. In the last period of the war, the absolute advantage in recruiting for military schools was given to candidates from among the Red Army and sergeants from the troops, and from among the latter, first of all, front–line soldiers were selected. At the end of the war, cadets who came out of civilian youth made up only 6.1% against 71% in January 1943 ^{[2, pp. 227-228]}. Therefore, among the recruits mobilized in the liberated regions of Ukraine and Belarus, few candidates for military schools were selected. In addition, there is evidence that there was a direct ban on recruiting people who had been under occupation to schools ^{[3, l. 74]}, however, it could hardly be strictly and fully observed. It should be added to this that since 1943, the terms of training in all Land Forces schools were increased to 1-2 years, and by 1945, the pre-war terms of training were mostly restored ^{[2, pp. 229-230]}. Thus, cadets from the liberated territories of Ukraine, Belarus, Moldova, who were admitted to military schools in 1944, actually had no chance to be at the front before the end of the war. The low number of cadets of these nationalities speaks for itself. The most striking example of this bias was Moldovans who were among the last to join the Red Army and "burdened" by mass ignorance of the Russian language, service in the Romanian army.

In conclusion, it should be noted that the consolidated albums of socio-demographic data of the Red Army roster have a very high information potential for statistical research on changes in the national composition of the Red Army during the Great Patriotic War. Further study of them may include determining the indicators of national representation in active and inactive troops, in various types and branches of the armed forces, as well as determining the absolute number and proportion of losses by nationality (albums contain information about the wounded and sick who were treated in hospitals). This will provide additional material for assessing the intensity of the use of representatives of various peoples of the USSR at the front and in the rear.

applicationThe national composition of the Red Army in four military categories (as of July 1, 1942 and January 1, 1945).

The percentages in the tables are indicated in relation to the number of each nationality in the Red Army

Data source: ^{[4, l. 194, 199]}; ^{[5, l. 1-17]}, see ^[1].Table P1.

Data for July 1 , 1942Nationalness

  Data as of January 1 , 1945