Hi-
Someone recently pointed out to me that some microfilms are of the census records kept at the National Archives, records that were hand copied from the originals. Because they were hand copied, there exists the chance of numerous mistakes. I was told to track down microfilms of the originals that are kept at the state level....not sure what is available for each state, but here in Wisconsin, the State Historical Society does have its own microfilm of the originals. I'm in the process of tracking those down to see if any of the info is different than what I have....but, I was told by someone who already did this, that I would definitely find mistakes. Just an idea for those that have found confusing things recorded for their family on a census.
-Cheryl (Feind) Janowiak
Hi,
I do believe that the LDS has microfilm of all the U.S. census records
from the original. However, the National Archives have the index (soundex)
for each state. Those were handwritten on cards, then put on microfilm. I'm
not so sure the LDS has all of those.
Barbara
Hello Cheryl and Everyone,
Always be alert for mistakes in any U. S. Census - even the originals!
Remember, the information was given verbally by one human (who might not
have known or remembered correctly), to another human (who may not have
heard or hand-written it correctly). Immediately, there is the possibly of
a memory problem, a foreign accent problem, a hearing problem, and a
penmanship problem.
And then we have the problems of humans who "transcribe" what was written in
the originals so we can have "indexes" that can be used by "search
engines"..... If the name starts with an "R" and the enumerator's
handwriting made it look like a "B" or a "K" or a "P" to the transcriber,
there's no way the search engine is going to find your name in a transcribed
list when you are searching for an "R", even if you use soundex.
Whether you are using a "transcribed" copy or microfilm of the original, you
must be flexible about the information, especially when it comes to
spellings and dates. All records are created by humans, and all humans make
mistakes. :-((
Carol Payne SnowWhite@Lakewebs.net
Jb: How true how true, but we all need to bear in mind what a blessing these indexes are in the end, whether they pertain to the censuses or anything beyond.
And it goes without saying, the CENSUSES themselves. Census records remain the bread and butter of American side genealogies, even if fraught with anomalies. In some ways these irregularities increase the overall "challenge", and makes each discovery more delightful (at least to me). Adds a little "spice" to the mix in the end. 
The same can probably be said for other English speaking nations (U.K., Canada, Australia, etc.).
On the other hand, I would have to imagine the percentage of errors would be lessened in more homogeneous populations, like that which would have been found in Germany and England years back. Not as many names would appear 'foreign' to the enumerators (except perhaps the Polish or Gallic). 