This needs further work, to fold in the concept of "Ephemeral Sources".
What is a "source"?
Most fields of study utilize the concept of "sourcing". Putting it simply, "sourcing" is the process of documenting where certain information came from. Traditionally, sources have been classified in terms of "primary", "secondary, and "tertiary" sources. This terminology is used in a wide variety of fields, and persons from one field understand fairly well what is meant when someone from another field refers to something as a "secondary source", etc.
In the BCG scheme of things, there are "Original Sources" and "Derivative Sources". This makes a certain amount of sense because it clearly distinguishes between the two main types of sources: the original records on which everything else is based, and the derivitive sources which are created from the original records (e.g, published compilations, extracts from original records, etc. This terminology is a whole lot more intuitive than that used by everyone else. The problem is, though, that everyone else uses a different set of terms. As useful as BCG nomenclature might be, it IS idiosyncratic. I'm not sure that adopting a terminology different from that accepted by everyone else, is very useful.
Many genealogists---especially those making extensive use of web resources---do not make extensive use of true primary sources. Usually the data they use is taken from secondary and tertiary sources. This is readily understandable: access to the original primary records---the handwritten county and church records, upon which everything else is based---is limited. To access them you'd physcially have to go to wherever those materials were being housed. Even then, securing access to the actual records is not assured. If you were doing this type of research professionally, than adherence to the use of primary sources would be expected. When people are doing this as a hobby (read: almost all genealogists), then the research burden of relying on true primary records is probably not reasonable. As a result greater reliance is placed on the use of what are technically secondary sources, e.g.:
Such approaches make information contained in primary records much more readily accessible. But there should be no confusion here: these are still secondary sources. Even a microfilm, or photocopy of an original record is in fact a derivitive source, and subject to the problems of information loss and error that are always contained in derivitive sources.
BCG notes that
The reality is that most genealogists will not be looking at original records, but will in fact make use of secondary sources. As a result, most source citations are going to be references to a secondary source. From this perspective making a distinction between primary and secondary sources seems to be an academic exercise with limited practical value. However, from other perspectives, the distinction is critical, and failing to make this distinction diminishes the value of a genealogical product. See The Importance of Sourcing
1. Original record. Traditionally, a "primary source" is an original record contemporary with a specific event. The person creating the record usually has immediate, first hand information about the event, or was at least a contemporary of those who told them about the event. Primary sources of this type include public records, letters, family bibles, contemporary newspaper articles, etc. They also include personal reminiscences of persons directly involved in the events, or were othewise eyewitnesses to the events. The Board for Certification of Genealogists prefers to use the term "original record" to indicate sources of this type. Others prefer to use the more commonly encountered "primary source"
2. Primary Source. Access to most original sources of the type described above, is often limited. Few persons ever see the original records describing a specific event; In most cases we are dependent on transcriptions or photocopies of the original materials. A will, for example, may be found in the county archives, but what is usually available to most people, is a transcription of the will. Transcriptions may be more or less accurate, but they are not the same thing as the original primary source. Most people have to rely on things like transcrptions of the originals. As long as what is provided is a verbatim copy, or at least an honest extract of what is found in the original, most people would accept it as the equivlent of a primary source. Here, a "Primary Source", is understood to be a copy of the "Original Record", seems like a reasonable way to distinguish such records from a true primary source.
3. Secondary Source. A secondary source is a source of information that is not contemporary with the event. The author has no direct or immediate knowledge of the event, and knows of it only by what he has been told, or read. He may have consulted a primary source, or an indirect primary source for his information. If
his own work may be elevated to the level of an indirect primary source. If he does not indicate his sources, or does so only in a general way (ie, "various county, state, and church records"), or gives a list of specific sources used (ie, Smith, 1911; Jones, 1848) without indicating the exact information that came from each source, his work is regarded as a secondary source.
4. Tertiary Source. A derivitive work, largely or completely making use of secondary sources