On 2011/02/06 18:54, John Laflin wrote: > 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. Your premise is extremely unlikely for the bulk of the census images. They are simply too dirty and faded for OCR to be of much use. Unless that is, you think OCR stands for Overseas Correctors and Readers. OCR relies on being told what the shape of each letter looks like. There is simply too much variation in handwriting for OCR to work. The bulk of the problems, both on Ancestry and FMP, stem from the fact that the transcribers were not native English speakers. Hence the sometimes laughable mis-transcriptions, like Somalia for Somerset. The number 1 rule of any transcription process is: Write What You See! If your first language is not English, then I can easily see, from examination of the images, how some of the errors have been made. -- Regards, Mike Fry Johannesburg