Thanks for everyone's replies. I'll make a few comments.
1. The legal database (more properly, collection of legal journals) I was using when the question came up was HeinOnLine. In my small city, it is available in one public library and two university libraries. I'm sure the access is similar in other cities. Similar resources like Lexis are also widely available in public libraries. So I don't think availability is an issue.
2. Jay is quite correct to discourage weasel words like "probably". For one thing, it prevents objective verification.
3. Sarah points out the following text from WP:NOR : "anyone--without specialist knowledge--who reads the primary source should be able to verify that the Wikipedia passage agrees with the primary source. Any interpretation of primary source material requires a secondary source." That seems to be broken. Examples of specialist knowledge which might be required are the ability to read a foreign language and the ability to understand mathematical notation.
Someone who can read music should be able to report from a musical score that it is in E-flat, even though that requires specialist knowledge. What the policy *should* require (somehow) is that anyone who can read music will agree that the score is in E-flat. The fundamental skills of the field should be assumed, and the policy should reflect that, imo.
4. Sarah wrote: "We use writers as sources, not databases and libraries." Nobody suggested libraries. I don't see that databases are excluded by any existing policy, provided that the process of extraction of the information from the database is verifiable.
Suppose I have a book about a serial killer, which lists all the victims one by one. I think it is perfectly ok to write "all the victims were women" after looking up each case in the book. It comes under "research that consists of collecting and organizing information from existing primary and/or secondary sources is, of course, strongly encouraged."(WP:NOR) I can't see how that is different *in principle* from reporting that all the articles on a particular subject in a particular database give the same story about something, provided that that observation is one that anyone can verify. Of course this criterion might not always be satisfied, but that shouldn't eliminate the cases where it is.
Zero.
__________________________________________________ Do You Yahoo!? Tired of spam? Yahoo! Mail has the best spam protection around http://mail.yahoo.com