Transistor matching method ?

[Deathlehem - experience the joy]

>From: Joachim Verghese <jocke at netcontrol.fi>
>Sometime back I did some tests on a batch of 20 2N3906s (all from
>one manufacturer). I forced a known constant current into the
>collector, and then measured Vbe and Ib. I also tried my best to
>keep the temperature constant. Here are the values for some of the
>transistors:
>
>>From the above figures one could conclude that the two parameters
>*are* correlated, but there seem to exist individuals with a hfe
>higher than expected. I also tested some 2N3904s, and the results
>were similar.

Could anyone tell me the procedure for matching transistors by hand, if
there is any more to it, that what is listed above?  I don't know what
values of each I'm ultimately looking for.  Do I correctly presume that
these will be listed in a transistor databook?  If so, which value headings?
The two above?

Once I've got this information, I'll lay it out nicely and put it into 'The
Eventual Builder's Guide' on my homepage, for everyone to use.

Thanks much,
Confusedly,
Ric
Ric Miller - founder of Deathlehem
  Wave files for ARF, Bernie and
     Deathlehem are available
 at http://www.pangea.ca/~rmiller