Most people reference Gott's book when giving Confederate strength at Fort Donelson. Gott mostly uses the "tabular statement" compiled at the time:
He then proceeds to make a few imputations for units not included above. Investigation has shown that every unit he imputed is already in this list. They are:
Culbertson's Battery of 300; these were the men manning the water battery, but were detachments from units in the list. The battery was manned by Maury's (Ross') battery, Coy A of 30th TN and Coy A of 50th TN. These units are on the list, and Gott double counts them.
Melton's scouts are listed in the table as having 15 men. Gott gives them 58.
Major Fielding Gowan's Tennessee cavalry squadron is listed on the table as having 60. Gott estimates 170.
The Kentucky cavalry coys were attached to Forrest's regiment, and are included in it (see the returns below). Gott doesn't list sources, but gives Huey's coy an incredible 112.
Also, for no reason Gott added 150 surrendered to the 48th TN.
Finally, there is an addition error in his artillery table.
We also have the returns for the formations a mere two weeks prior to Fort Donelson:
Of these formations, the majority of the 4th Division, the whole of Floyd's "division" and Clark's brigades, and the artillery and 7 regiments of Buckner's division were at Donelson. Fortunately Buckner broke down the regiments strengths in his report and it is close to 7/12ths of his January return, and can be accepted. The PFD at Donelson can be (over)estimated thus:
Thus the estimate of 13,000 given by the likes of Pillow seems accurate. Note that the highest figure given by any confederate is by Preston Johnston, but he double counted Clark's and Floyd's brigades. Removing the double counts give 15,000, which is consistent with the returns.