How accurate is log referrer data?

[count_zer0]

I am in the process of analysing a client's web stats. I have been surprised by the relatively small number of SE referrals in the logs:

9297 Referred by search engines: 0.36%
- Unique search engines: 53
- Unique query terms: 4438
- Unique keywords: 5650
- Keyword count: 15361
249841 Referred by external URLs: 9.73%
- Unique URLs: 47349
947194 Referred by local URLs: 36.89%
- Unique URLs: 1777
1361082Unknown referrer, direct from bookmarks, robots, and typed URLs: 53.01%
=======
2567414(+ excluded targets: 5252323 + excluded referrers: 18 + corrupted lines: 672 = 7820427)

This is a month's logs.

These figures were obtained using a script called Relax which has accurate definitions for 400+ search engine referrers. I have cross-referenced them with Analog and they tally.

What I want to know is:
1. Is this an unusually small number of SE referrals?
2. Is it usual to have 50%+ of "unknown referrer"s?

I'm stumped. Does anyone have any wisdom to impart?

[FrenchGuy]

1. I have the experience that log analyser do not alway recognize SE referer as such but put them as "other external URL". I work with websuccess, which is a wonderful log analyser, but it recognize just 20% of all SE referer as such... I have to pick up the other SE referer in the details of "other external referer" which fortunatly, are available as excel sheet...

2. In your case, "Unknown referer" include direct access to your site (people type your domain or have it as a bookmark, pass each other per Email etc. Last but not least: spiders are also included here. Their access percentage varies from site to site but can make out a big part (I think about 7% for my sites)