Hi John The census were transcribed by humans for both Ancestry and FMP Many other directories and printed sources were OCRd but the census were not as the OCR software is still not available that could make sense of the many different styles of handwriting This came as a reply to may enquiry to Ancestry some years back And yes FMP and Ancestry use different transcriptions The most widely used transcript is the 1881 which is still used by many even though they have their own images Nivard Ovington in Cornwall (UK) > Recent comments on this refer to errors by the Ancestry transcribers. It is > important to remember they were not transcribed, but produced by OCR. It is > not easy for this software to correctly interpret the 'flowery' Victorian > handwriting and of course impossible to make judgments and interpretations > on what is written. > > Hence with my own name. Quite frequently in the original entry, the > horizontal stoke across the middle vertical stoke of the letter 'f' extends > and cuts across into the vertical stoke of the letter 'l'. Now whereas a > human subscriber would recognise this and see an 'l', to the OCR software > the 'l' has become a 't'. Hence the name in Ancestry is not Laflin, but > Laftin. This occurs quite frequently as a quickly written 'a' can be taken > as an 'o' making the entry Loflin' and a flowery L with stokes at the top > and bottom turn the Ancestry entry into Saflin. > > To my knowledge only the 1851 and 1881 census returns have human > transcriptions by FHS's and the LDS and are therefore much more reliable. > > It is also clear the Ancestry and Find my Past use different OCR software, > as an 'error' on one can be correctly interpreted on the other. > > John Laflin