The definitions of primary and secondary sources often given by family historians are less rigorous than those used in other disciplines. The more rigorous approach is taken here.
Primary sources are those where the information is supplied directly by the persons concerned. For example, at the marriage ceremony, the bride and groom supply their details for the church and/or civil registers and sign their names. Both of these marriage registers are primary sources of the information. In this particular example, the index to the South Australian Marriages Registrations 1842 to 1916 (South Australian Genealogy & Heraldry Society, 2001) was transcribed from the civil certificates and is therefore a secondary source.
The series of SAGHS indexes to civil birth, marriage and death registrations were intended to be accurate indexes to the certificates, as written, including any errors on the originals. Some further errors were inevitably introduced by SAGHS due to the difficulties of interpreting some of the handwriting on the certificates. Transcription and data entry errors were kept to a minimum by checking the work at each stage.
FamilyHistorySA's database of early marriages has mixed level sources. Those prior to civil registration are mostly taken directly from the church registers and are therefore secondary; those based on the civil certificates but actually extracted from the SAGHS index are therefore tertiary (3rd level). Some of our records contain extra information such as additional given names and previously married women's maiden names. When obtained from the SAGHS birth index, the extra information is still tertiary but when from the Biographical Index of South Australians 1836-1885 (SAGHS, 1986), it becomes quaternary (4th level) or lower.
The primary sources of our passenger lists were the original applications for passage made to the shipping agents. These were transcribed into a register of those wishing to emigrate, a secondary source, prior to being destroyed. They were then assembled into proposed passenger lists that resulted in certificates of departure after the passengers were checked off before departure of the vessels, and finally certificates of arrival. These official lists were the shipping companys' primary sources of their passengers. But were they primary sources of the information on each passenger or family? This depended on how the check was carried out. If the details were checked with the passengers, then the listed information became primary. Otherwise, the official list is a tertiary or lower level source of information about each passenger. It is usually not known how this was done. In many cases, the official lists have been lost and only the manifests remain. These lists are secondary or lower as they are transcriptions of the names on the official lists onto the cargo manifests. Newspaper passenger lists are lower level still. Each transcription has the potential for omissions and the introduction of new errors. Newspaper lists have a notorious reputation for their high error rates. However, they can't be ignored because they are often the only surviving records of the arrivals.
We have just seen that for sources resulting from complete or partial transcription from one level to the next lower level generally have worse accuracy at each stage. So, for this type of source, the historian needs to access the highest level that is available — ideally the primary level. One must bear in mind, though, that primary sources are generally not 100% accurate.
Take the sources of the Register of Emigrant Labourers Applying for a Free Passage to South Australia 1836-1841 (PRO CO 386/149-151), for example. Each applicant verbally provided their name and details to the shipping agent. The majority were illiterate, so the shipping agent, who may have been only semiliterate himself, had to spell the names as best he could. Also, many applicants lied about their ages and/or occupations in an attempt to increase their chances of being accepted. Hence, the shipping agents' lists were nowhere near 100% accurate and any information obtained from the register must be backed up by other independent sources. The original shipping agents' lists have been destroyed.
Consider the accuracy of the information given about a particular event in a person's life by the person concerned, at different times throughout life. The source is always primary but the information may well vary from time to time. Usually it becomes less accurate as time passes and memory fades, although this is not always the case. A controversial event may have been covered up to some extent when originally reported. Later in life the person may not have felt the need to hide or vary the details.
So, what about composite sources? Here, the family historian needs to take great care. The majority are poorly researched and therefore result in a further degradation of accuracy. However, by contrast, thoroughly researched composite sources usually result in improved accuracy. In some cases, the accuracy of the resulting composite information can be higher than that of the individual primary and other sources! This is because comparing numerous sources often enables the resolution of spelling errors, name variations, dates and other details.