The observations and questions being made about the various methods used to measure and record different aspects of the frames listed in the Bike Geometry Project's listings are all excellent.
Perhaps I should toss out some caveats before things get too far down the path. In short, treat all of the data there as suspect in the particulars.
1) There is little to no standardization of the tube lengths. In general, I think CS is taken as center of BB (or spindle) to center of axel, but not always. "Size", which is a marketing term is conflated with "seat tube length", which is a design term for which there at least 2 different meanings (c-c and c-t). And of course top tube length and effective TTL are conflated and generally are assumed to be c-c but who knows.
2) Most of the newer frames in the list are from catalog specs. Many of the older frames are (by necessity) from hand measurements. Even if you assume that all specs were correct (they aren't) and that all submitters have measured the frames the same ways (they haven't), we need to understand that these are fundementally different kinds of data.
So, the data in the spreadsheet needs to be understood as being approximations in all aspects. This begs the question then, does it have any value?
I think it does for 2 reasons. First, it's a starting point and better than nothing.
Second, approximations are enough to help support certain kinds of reasoning. For example, I've been doing a bit of reading (for work) in the area of library sciences. One of the ideas here is the distinction between CATEGORIES and CLASSIFICATIONS. In classifications, you have non-ambigous boundaries between classes and all members of a class are equally representative of the class. (e.g. birds or cats) In categories, you have ambigous boundaries between categories and some members are more representative of the category. (e.g. bluegrass or jazz).
I think bike frame designs can be thought of as categories in this sense and when you do, approximate data works just fine. Should a bike be categorized one way or another based on a 5mm difference in CS length? Or base on a 0.5 degree difference in HA? You and I are free to decide and to decide differently. What's astounding though is that in the midst of this certainty we can both say "Jazz" or "Road Racer" and more or less communicate with each other. My hope is that the Bike Geometry Project helps foster such discussions about the classics.
Here's a recent example. Thanks to the kind input of a CR member over the weekend, we've learned that the 1970 - 1975 Gitane Super Corsas and Tour de France models all shared the same geometry. (72.5 parallel, Rake: 5.5: CS: 45). This is almost the same as the first Trek sport touring bikes. And other than the longer CS, very close to the 75 Fuji America.
I think making those connections is very cool.
I believe the old Raliegh Clubman's were 72 parallel designs. It will be cool to get some of their rakes and CS figures documented. Lots of 72 parallel designs on the market again. Distant echos...
-Dave Mann, Boston, MA
-------------------------
THE BIKE GEOMETRY PROJECT
A community effort to document and compare bike geometries
http://home.comcast.net/