Sherman's returns show a steady diminution of his available numbers during July and August, though, as he himself has said, it was not altogether from casualties on the battlefield and the diseases of the camp. [Footnote: Memoirs, vol. ii. p. 134.] The term of service of all the troops enlisted in the spring and summer of 1861 for three years was now ended, and an interval occurred in which the new levies under the law to enforce the draft had not yet reached the field, and the army was depleted by the return home of the regiments which had not "veteranized" in the last winter. He had present for duty, on July 31st, 91,675 officers and men; on August 31st, 81,758. Sherman's statement of his losses in battle and his comparison of them with his opponents is a model of candor and fairness. With the light we now have, he might properly have increased considerably his estimate of Johnston's casualties. [Footnote: Memoirs, vol. ii. pp. 131-136.] General Hood was quite right in arguing, in his memoirs, that the wounded in a campaign are not all a permanent loss to an army, "since almost all the slightly wounded, proud of their scars, soon return to the ranks." [Footnote: Advance and Retreat, p. 217.] But what I have said above shows that he was entirely astray when he concluded that the difference in the returns of his effective force at the beginning and end of the campaign would show the number of killed and permanently disabled. The absence of data as to the additions to his field force through the means which I have analyzed, shows how absurd a result was drawn from his premises. The reports of casualties are not unfrequently faulty, but with all their faults they would be much more valuable if a complete series existed which could be compared and tested. It would require a minute examination of all returns, from companies to divisions, to determine accurately how many men returned to duty after being wounded or captured. The imperfect state of the Confederate archives would prevent this, if it were otherwise practicable. The statistical returns are conclusive for what they actually give, but inferences from them must be drawn with care. As an illustration (in addition to those already given) it may be noted that the Confederate cavalry made no returns of casualties or losses, and they do not appear at all in the Medical Director's report which General Hood makes the basis of his own assertions. [Footnote: Official Records, vol. xxxviii. pt. iii. p. 687.] How grave an omission this is will be partly seen from the fact that Wheeler's corps, which reported 8000 men present for duty on August 1st (the last return made), was in such condition when he reached Tuscumbia after the raid in the rear of Sherman's army, that its adjutant-general doubted if more than 1000 men could be got together. [Footnote: Letter of General Forrest to General Taylor, Sept. 20, 1864, Official Records, vol. xxxix. pt. ii. p. 859.]