I have read multiple books which stated that when the battle commenced there were less than 30,000 Confederates on the field vs 85,000 Union soldiers. In the afternoon, reinforcements started to arrive after having hurriedly marched from Harper's Ferry.
A quick google search showed 87,000 Union vs 38,000 CS (Wiki) and another site showed 87,000 Union vs 45,000 CS (battlefields.org).
Okay, thanks, so the 87,000 is the number of Union soldiers on the field at the start of the battle. Let's have a look at that number.
The number is directly from McClellan's report, which gives 87,164 as such:
1st Corps 14,856
2nd Corps 18,813
5th Corps (minus one division) 12,930
6th Corps 12,300
9th Corps 13,819
12th Corps 10,126
Cavalry Division 4,320
Now, immediately the problem should be clear - this is not a state for the morning of the 17th. It is a list of all forces who were present in the battle at all.
But worse is that these numbers are
not the number of troops actually on the field - they're something else. 1st Corps only took about 9,000 men into battle, and indeed on the 18th Meade reported his strength in officers and men present for duty as 6,364 - it's a few days later that his strength is reported much higher, after stragglers have closed up, and indeed it looks like the strength reports above are more or less the September 22 states plus reported casualties.
So the list from McClellan's report is the strength of all forces who were in the battle,
before straggling. (ED: I suspect it's his PFD strength from the missing 10 September report.)
Is the same true of Lee's strength?
Well, no. Not at all. The 38,000 number is based on Carman, who gave the breakdown as 3,629 artillery, 4,500 cavalry and 29,222 infantry. Note that this is "engaged" infantry, not all infantry - Carman explicitly disregarded Armistead's brigade and two brigades of AP Hill as they were not "engaged".
Since only AP Hill's division arrived late in the day, I'll disregard them but look at all other Confederate units.
McLaws - this is estimated based off McLaws' AAR, and is infantry actually present on the field (
2,823 infantry).
RH Anderson's division - Carman bases this estimate off DH Hill saying that Anderson brought 3000-4000 infantry to support him. Since Armistead was detached he is not counted here, and was not counted where he actually was either. Anderson's division was 5,324 PFD on the 22nd, and suffered about 1,300 casualties at Antietam - it's hard to justify lower than
5,000 on the field counting Armistead.
DH Hill claimed he had 3,000 infantry, but Carman refutes this and calculates 5,449 infantry based off reports - this has interesting implications for RH Anderson's division, as this size is based off a report from DH Hill. In fact it looks like Carman did not go far enough:
Ripley's brigade, Carman determined from correspondence that there were 1,349 officers and men in action.
Rodes' brigade, Carman accepted Rodes' claims of less than 800 but in fact three of the five regiments (in correspondence with Carman) claimed numbers which totalled 840. This suggests that Rodes' brigade was larger than Carman thought, perhaps up to 1,400 strong.
Colquitt's brigade - adding up regimental strengths and accepting Carman's guess for the 13th Alabama gives about 1,800 or so.
Garland's brigade is given 756 by Caman, but the 5th NC alone had 625 men at Antietam and there were four other regiments in the brigade. An estimate of 1,800 seems a safe lowball, on September 2 they had over 2,500 PFD
GB Anderson's brigade was given by Carman as 1,174
Thus a reasonable total for DH Hill's brigade is more like
6,500 infantry.
Walker's division is given by Carman at
3,764.
DR Jones' division is another one which is problematic. Carman estimates 3,311 infantry, while going by brigade:
Toombs' brigade was estimated as being 357 in the 2nd and 20th Georgia, 281 in the 15th and 17th Georgia, and 140 men present in the 11th Georgia (of which only 5 companies were present). Total 781.
Garnett's brigade is given as only 261 officers and men.
Kemper's brigade was estimated by adding together regimental reports and estimating the 28th Virginia (to be 443), but others reported the brigade at 500-600 muskets which implies that the 58th was a strong regiment. Use 500 here.
GT Anderson was actually under someone else at this point, Carman estimates 597.
Drayton was estimated by Carman at 465, excluding the 11th Georgia (detached), but he used 200 for the 15th SC when he'd been told by a veteran that the number should be 300. He also did not include the Phillips' Legion, which was 135 men ("To Honour These Men") and thus the total should be 700 (465, plus the 100 dropped from the 15th SC, plus 135 in the Legion).
Jenkins' brigade - Carman quoted a newspaper for 755, but the brigadier reported 1,250 exclusive of the 1st SC (at 106). So 1,356.
So the total actually with Jones was
3,600, with one brigade elsewhere.
Evans was where GT Anderson's brigade had gone.
Hood claimed about 2,000 men in both brigades.
Evans is someone for whom Carman reports 284 infantry.
And as before
GT Anderson is 597.
So this is about
2,880 infantry.
Jackson is given by Carman as 1,784 infantry, but as I've already mentioned this appears to be the state as of arrival on the field, with Starkes claiming 1,400-1,500 in his brigade actually in action.
Taking Starkes' brigade as being about 1,400 on the morning of the 17th and expanding the other brigades pro-rata gives about
4,400.
Ewell was claimed by Carman as
3,904 infantry, and I'll accept it despite some irregularities.
And
AP Hill had not arrived yet.
Total infantry count: 32,871 on the field on the morning of the 17th, minimizing Confederate size in many cases.
Plus the artillery (-337 for the artillery with AP Hill) and cavalry, this gives about 40,500, based either off Carman, or Carman's sources, or in places direct sources Carman rejected.
But, you may claim, isn't this still much weaker than McClellan? Well, yes, but McClellan's strength is
before straggling while the strength as above is
after straggling. With straggling so bad (more than a third of Meade's division didn't show up to fight, and some corps seem to have been even worse) the two numbers are simply not comparable.
So what is?
Well, Carman of course! Carman also calculated numbers for the Federals, and he estimated 46,146 Federal infantry engaged at Antietam. This has the opposite problem to the Rebels that in some cases Carman used PFD strength to calculate Engaged (which means the strengths of 2nd and 9th Corps are a bit dubious) but we'll use it for now, and we also need to add in most of 5th and 6th Corps as they were not Engaged.
5th Corps can be estimated at about 7,000 infantry (most of which is Morell at 5,407 PFD, with Buchanan and Lovell's brigades assumed to be about 2,500 between them and the 2nd and 10th NY very small) and 6th Corps was about 9,000 infantry if no straggling took place (Battles and Losses II p 595 holds that 75%-80% of 6th Corps were combatants). Since Irwin's Brigade
was engaged and was about 1,700 (Carman) along with Buchanan's brigade the same at about 1,600, this means about an extra 12,700 to Carman's number.
Thus McClellan's infantry present, counting everyone, is not more than 59,000, and infantry to infantry on the morning of the 17th it's
32,000 Confederate to 59,000 Union.
Except there's another problem. See, some of those hadn't arrived yet when the battle starts - at about 11 AM Morell is just closing up and Franklin's column is entering Keedysville.
Cutting them out again means that the troops actually available to McClellan in the morning phase are... pretty much Carman's estimated engaged, plus Lovell's brigade but minus Irwin's, so it's a wash.
Thus, in infantry it's
32,000 Confederate to 46,000 Union. About a 3:2 advantage until Morell and Franklin arrive, whereupon it becomes about a 2:1 advantage until AP Hill shows up.
This is what happens if you measure both sides in the same way, and you're careful about it.
Taking whole campaign strengths instead you can get it much closer, with Confederate strengths around 78,000 (that's October 10 plus add-back of casualties) and Union strength more like 110,000 if you get really liberal with who you include. (September 20th report ~93,000 PFD, plus 13,000 casualties, this includes Couch.)