On Tue, Jan 06, 2009 at 11:04:39AM -0800, Delirium wrote:
But I'd also be skeptical of a general mathematical article, on something like [[calculus]] or [[statistics]], which was constructed mostly from journal articles.
Of course. Articles like [[Calculus]] are sufficiently elementary that they can be (and universally are) written almost exclusively from textbooks. Moreover, any result that might be of interest but is not in elementary textbooks would probably not be included in these articles because of their scope. Such results would be in more specialized articles. These are the articles that I am concerned with in this thread. They are also, I believe, the articles Phil is concerned with.
Apart from sources cited only to maintain the historical record, the main reasons for citing primary sources in the math articles I have worked on are:
* For results that are of sufficient encyclopedic interest to be included, but not of sufficient interest that textbooks cover them. These often unproblematic because the claims made in the article are just brief summaries of results.
* For results that are sufficiently new that they are not yet covered by secondary literature. I can think of several research programs with numerous papers by numerous independent authors, with significant scientific interest, but no coverage outside of journals. These areas have no sources that on their face are accessible to a reader without specialized knowledge; the wikipedia article may be the most accesible writing on the subject.
- Carl