In my previous post I explained that historians have overcounted slaves in the New York census of 1830. Instead of the traditional count of 75 or 76, I estimated that there were in fact only between 38 and 51 slaves actually tallied on the census forms.

One reason for the latter-day counting error, evident in both the work of Carter Woodson in 1924 and in ancestry.com’s search engine results today, is what I have coined “ghost tallies” – 19th century cousins of the “hanging chad” that were common in census books.

The problem results from the census taker using an ink pen to record his results. Below is an example from a census page for Washington County, New York, in 1830.

Notice in the image above how the column 6th from the left appears to have a “1” mark in about the middle of the image. Ancestry.com reads this as a mark for a slave, held by a person named “Lunn House” or “Sunn Hoase” in Washington County, New York. But at the bottom of the page, the tally for this column is blank.

Looking closer, we can see similar marks that don’t contribute to the total at the bottom, and the difference between this “ghost tally” or “ghost mark” and the real marks becomes obvious. The real marks are more vertical, but still with a slight (/) rightward lean. The ghost tallies, meanwhile, are slight more horizontal ( – ) and aren’t quite as dark. There are at least three ghost tallies in the image above.

A number of things might explain this or the many other ghost tallies in the census. In some instances they appear to be ink marks that have bled over from other pages, either the page behind, on the previous page, folded over onto the current page, leaving an impression. Sensitive microfilm or modern scanning equipment might pick this up as more pronounced than it appears on the original page.

These ghost tallies not only influenced certain earlier attempts at counting persons in the census, but they also continue to confuse modern researchers who use ancestry.com for genealogical research.

Another “ghost tally” From Washington County, 1830.  The mark in the category of female slave between 10 to 24 is not counted in the column total on the bottom of the page, nor in the row total on the right edge of the page. Meanwhile, the mark for a free black male on the second row has been counted by the enumerator. Notice the many bled-through digits that appear angled in the other direction.