The actual enumerated number of slaveholders was 347,255. After that, a rather unique "scientific" method was used and "the number swelled to about 2,000,000." Further "academically-accepted" massaging of the numbers swelled the number to "about two million and a quarter."
SOURCE: Southern Pamphlets on Secession, November 1860 - April 1861, edited by Jon L. Wakelyn, University of North Carolina Press, 1996, ISBN: 0-8078-2278-7, p. 79
James D.B. DeBow, "The Interest in Slavery of the Southern Non-Slaveholder," (Charleston: Evans & Cogswell, 1860)
DeBow begins as follows, this text running continuous to start his pamphlet.
"When in charge of the national census office, several years since, I found that it had been stated by an abolition Senator from his seat, that the number of slaveholders in the South did not exceed 150,000. convinced that it was a gross misrepresentation of the facts, I caused a careful examination of the returns to be made, which fixed the actual number at 347,255, and communicated the information, by note, to Senator Cass, who read it in the Senate. I first called attention to the fact that the number embraced slaveholding families, and that to arrive at the actual number of persons which the census showed to a family. When this was done, the number swelled to about 2,000,000."Since these results were made public, I have had reason to think, that the separation of the schedules of the slave and the free, was calculated to lead to omissions of the single properties, and that on this account it would be safe to put the number of families at 370,000, and the number of actual slaveholders at about two million and a quarter.
"Assuming the published returns, however, to be correct, it will appear that one-half of the population of South Carolina, Mississippi, and Louisianna, excluding the cites, are slaveholders, and that one-third of the population of the entire south are similarly circumstanced. The average number of slaves is nine to each slave-holding family, and one-half of the whole number of such holders are in possession of less than five slaves."
free dixie,sw
Now, Watie will argue that the census is worthless and that one has to look at the tax records county by county, but he doesn't present any number that would be yielded by that method. What he actually asserts is that the number DeBow gives for enumerated slaveholders, comprising about 5-6% of the white population is basically correct. But he also believes that this number incorporates the families as well.
Again, in the absence of evidence beyond Watie's shouting "no it isn't" or occasionally citing the unpublished dissertation of a dead professor, these are the numbers. If you've got some other ones, let's see 'em.