It is important to realise that the census, like any other source, is full of errors and ambiguities, and it is helpful to know how there occur.

The census records where individuals were on census night, and it is possible that some people were missed. No arrangements were made to count itinerants, travellers and night-workers in 1841 and many may have been missed in later years. It is also possible that some households were entirely overlooked. In some cases the names of people are not recorded - as could happen at lodgings houses (especially houses "of ill repute") that did not record the names of their overnight guests.

It should also be remembered that many people could not write so the census enumerator could only guess at the spelling of a surname. If the householder came from a different part of the country, and had a strong regional accent, the spelling might be very different to the way it would have been spelt in the area where he was born.

The place of birth could cause difficulty. In addition to the census enumerator misunderstanding what was said there was the whole question of what place to record. In England it should have been the town or parish and county, but interpretation was very variable. Sometimes the place would be the name of a farm or a group of cottages, such as Doolittle (in Kings Langley). Someone born in the parishes of St Michael, St Peter or St Stephen may be recorded by the name of the parish or as St Albans. There can also be problems as to county - particularly for people born near a county boundary - which is compounded where the county boundary has moved. The same problem can occur with town and parish boundaries - and my grandfather was born in Sandridge Parish. A few years later the farm was within the City of St Albans (but still Sandridge Parish) and by the time of the 1901 census the farm was be in the parish of St Saviours, St Albans! I have come across cases where it appears that the census enumerator was given the place name and then added the wrong county. For example I have come across a case where Westwick (a farm and cottages the enumerator is unlikely to have heard of) in Leverstock Green is recorded as Westwick, Cambridgeshire (a village just outside Cambridge).

Ages could cause problems. Apart from the inbuilt uncertainty in 1841 it seems that some husbands did not want to admit that they were younger than their wives. In other cases the ages were probably estimates - and the excess number of people with exact decade ages such as 40, 45, 50, etc., compared with intermediate ages in the 1881 census reflects the high level of guessed ages rather than peaks in the birth rate at five year intervals.

Having collected the household schedules and put them in order the enumerator then had to transfer the data to the census books - with all the problems associated with this. Where the householder had filled in the schedule the enumerator might not be able to read the writing - and in any case there was plenty of scope for errors. Many enumerators clearly found this a tiring task - as often the quality (and probably accuracy) of the script deteriorates the further through the book you get. They also took liberties and as most people in a town would have been born in the town they may have abbreviated the name so that in 1881 Bishops Stortford became "Bps Stfd" in one enumerator's book.

The book was then used to gather the statistics - and additional notes were made - for instance the words "Domestic Servant" or DS might be added to an occupation to distinguish between a groom who looked after the horses in a private house from one who worked in a commercial stables (ND for non-domestic). A check mark was also made to show that the return had been counted - and this is quite often positioned so that it partly obscures the age - and some unexpected ages arise because the obscured digit has been misread.

The books then went into storage and some may have become lost or damaged (as far as I know there are no major problems in Hertfordshire). More insidious is that in many cases the writing has faded. This can become even worse when the books are filmed and, for example, parts of the Tring 1841 census returns are unreadable on the microfilm.

There is then the question of reading the microfilm. If the person who is reading the film is experienced in reading the hand writing at the time, and has a good idea of the local surnames, place names and occupations they are likely to encounter - the error rate will be low. [One of the objections to the initial indexing of the 1901 census is that the people carrying out the indexing may not have the appropriate skills.] One common type of error is that many of the capital letters were very much OTT by present day standards and can be hard to recognise. Recently I was looking through a microfilm and came across someone who appeared to have the given name of "Miller" I then looked again and decided that it was almost certainly "William". The reason of this initial mis-read? The "ill" was no problem to read but the swirls that made up the capital letter were hard to untangle. But surely, you might say, it is not possible to distinguish "er" from "iam". The trouble was that the enumerator was tired and in a hurry and that the final group of letters simply degenerated into an almost horizontal line with no details.