The 1881 Census of the Wirksworth Area (17,469 entries) appearing on the CD sold by the LDS church was checked against microfilm of the Census bought from the PRO.
At the same time, the 1881 Census already appearing on this website since Sep 1999 was checked against the same microfilm.
The new 1881 transcription was also checked against existing transcriptions by John Palmer for 1841-51-61-71 and 1891.
The results are given below.
The LDS CD was found to have a surprisingly large number of errors. These were mainly in:
This is usually written only once in the Census, the rest of the family being written "ditto". So a mistake in reading the surname condemns the rest of the family to being in error too.
A good "nose" for locally unlikely surnames is essential for a good transcription. The strange surname can then be checked against other Censuses in database, searching on firstname and parts of the surname that appear certain. Surname errors are usually "fatal" for the transcription user who cannot run searches. Surname errors found in the LDS transcription are listed. A total of 866 surname errors were found.
The 1881 Census suffers from the attentions of statisticians, who were in the habit of crossing through ages, making 1/7, 2/3, 3/5, 7/9 etc difficult to distinguish. Once again, comparison with other Census years via database proved a tremendous help. Accurate ages are perhaps the best way of separating people in the Census with the same names. A total of 240 age errors were found.
PLACEs of BIRTH.
A good knowledge of placenames of Derbyshire is essential here, and a big Atlas Index or Internet database of English placenames are sometimes very useful. Many errors were found, but perhaps they were not "fatal".
Words in a different hand are often found added to the Enumerator's. These later additions are considered to be by a statistician, attempting to classify the occupation under a heading available to him. As such they have been omitted as they contribute no extra information useful to my readers.
Transcriptions of other Census years on database were the greatest possible help, as these allowed searches for part names, and comparison of age and place of birth.
The impression gained by John Palmer is that the LDS transcribers had a poor local knowledge of Surnames and Places.
Many surname transcriptions were frankly ridiculous, and these
should have been picked up by LDS re-checking but were not. It is not sufficient for the LDS to remark on difficult handwriting or warn its customers to check the original document.
METHOD of CHECKING
The data in the CD was exported to Word Processor, and reformatted to match the layout in the Microfilm. The file of 18,000 entries was split into 36 Enumeration Districts.
The microfilm was cut into strips of 3 frames and inserted into microfiche jackets to produce "pseudo-fiche" that could be used easily on my microfiche reader, and also grouped in Enumeration Districts.
The WP files were checked against the Fiche screen at the rate of 8 frames (200 entries) per day. Suspicious CD transcriptions were checked against the author's combined database of the other censuses: 1841-71 and 1891. Corrections found were edited into the WP files.
On completion of an ED, the edited WP file was exported back into database, and typos found and corrected by indexing all fields. On completion of all EDs, the whole Area was encoded by DPL, processed to website format, exported to WP, converted to ASCII, renamed as HTM and uploaded to website. A new hyperlinked Index was generated and uploaded to website.