In an earlier and lengthy post on this blog about how to use the newly released 1940 Census to search for Chinese immigrants, about halfway through the entry under the heading, "Some Strange Findings" I noted that I had found a few instances of a serious over-count of the number of Chinese in a few communities where I was searching for some other purpose.
As noted in that earlier post, the Census uses classification codes for recording information during face to face visits to households. Thus, in recording the race of the respondent, a code of C2 stood for Colored and C4 for Chinese. However, I found several instances where the count given for Chinese did not seem valid. Checking the original census record sheets online, I discovered that the codes had been accidentally reversed, and C2's got counted as Chinese and C4's as Colored, thus inflating the number of Chinese.
This problem did not begin with the 1940 Census as I also recently found the same error had occurred in a much earlier census. The 1910 Census listing for Summerville, Georgia, shown below indicated there were 21 Chinese (only labelled as C on this summary table, but the names in the list do not 'appear' to be Chinese. An online check of the actual record sheets confirmed that none of the 21 were Chinese but were Colored.
I did not check samples from the 1920 or 1930 census, but given that this over count of Chinese occurred in some locales for 1910 and 1940, it is reasonable to assume that the same error may have occurred in those intervening Censuses, and possibly even in some prior to 1910. I only found the tip of the iceberg, but the exact size of the unseen part of the iceberg remains to be determined.