<P> In 2013, Jankowski wrote that since the beginning of the war, French army units had produced états numériques des pertes every five days for the Bureau of Personnel at GQG . The health service at the Ministry of War received daily counts of wounded taken in by hospitals and other services but casualty data was dispersed among regimental depots, GQG, the État Civil, which recorded deaths, the Service de Santé, which counted injuries and illnesses and the Renseignements aux Familles, which communicated with next of kin . Regimental depots were ordered to keep fiches de position to record losses continuously and the Première Bureau of GQG began to compare the five - day field reports with the records of hospital admissions . The new system was used to calculate losses since August 1914, which took several months but the system had become established by February 1916 . The états numériques des pertes were used to calculate casualty figures published in the Journal Officiel, the French Official History and other publications . </P> <P> The German armies compiled Verlustlisten every ten days, which were published by the Reichsarchiv in the deutsches Jahrbuch of 1924--1925 . German medical units kept detailed records of medical treatment at the front and in hospital and in 1923 the Zentral Nachweiseamt published an amended edition of the lists produced during the war, incorporating medical service data not in the Verlustlisten . Monthly figures of wounded and ill servicemen that were treated were published in 1934 in the Sanitätsbericht . Using such sources for comparisons of losses during a battle is difficult, because the information recorded losses over time, rather than place . Losses calculated for particular battles could be inconsistent, as in the Statistics of the Military Effort of the British Empire during the Great War 1914--1920 (1922). In the early 1920s, Louis Marin reported to the Chamber of Deputies but could not give figures per battle, except for some by using numerical reports from the armies, which were unreliable unless reconciled with the system established in 1916 . </P> <P> Some French data excluded those lightly wounded but some did not . In April 1917, GQG required that the états numériques des pertes discriminate between the lightly wounded, treated at the front over a period of 20--30 days and severely wounded evacuated to hospitals . Uncertainty over the criteria had not been resolved before the war ended, Verlustlisten excluded lightly wounded and the Zentral Nachweiseamt records included them . Churchill revised German statistics, by adding 2 percent for unrecorded wounded in The World Crisis, written in the 1920s and the British official historian added 30 percent . For the Battle of Verdun, the Sanitätsbericht contained incomplete data for the Verdun area, did not define "wounded" and the 5th Army field reports exclude them . The Marin Report and Service de Santé covered different periods but included lightly wounded . Churchill used a Reichsarchiv figure of 428,000 casualties and took a figure of 532,500 casualties from the Marin Report, for March to June and November to December 1916, for all the Western Front . </P> <P> The états numériques des pertes give French losses in a range from 348,000 to 378,000 and in 1930, Wendt recorded French Second Army and German 5th Army casualties of 362,000 and 336,831 respectively, from 21 February to 20 December, not taking account of the inclusion or exclusion of lightly wounded . In 2006, McRandle and Quirk used the Sanitätsbericht to adjust the Verlustlisten by an increase of c. 11 percent, which gave a total of 373,882 German casualties, compared to the French Official History record by 20 December 1916, of 373,231 French losses . A German record from the Sanitätsbericht, which explicitly excluded lightly wounded, compared German losses at Verdun in 1916, which averaged 37.7 casualties for each 1,000 men, with the 9th Army in Poland 1914 average of 48.1 per 1,000, the 11th Army average in Galicia 1915 of 52.4 per 1,000 men, the 1st Army Somme 1916 average of 54.7 per 1,000 and the 2nd Army average on the Somme of 39.1 per 1,000 men . Jankowski estimated an equivalent figure for the French Second Army of 40.9 men per 1,000, including lightly wounded . With a c. 11 percent adjustment to the German figure of 37.7 per 1,000 to include lightly wounded, following the views of McRandle and Quirk, the loss rate is analogous to the estimate for French casualties . </P>

German troops almost reached paris during the battle of the argonne forest