I read Kris Carr's message regarding the use of a formula for allocating
funds for serials with considerable interest, because I was about to
post the same question to the list myself.  Two recent articles (one by
Charles Lowry and another by L. Rein et al.) present  convincing arguments
in favour of allocating funds for serials as well as monographs.  At my
institution, we have studied the feasibility of applying such a formula, and
have come up against what seem to be insurmountable difficulties.

Specifically:

1) Reliable figures for the universe of serial publications are hard to
obtain, particularly for interdisciplinary subjects.  Too many arbitrary
decisions are required.

2) Allocating funds for titles which are important for more than one
department is a problem.  Is prorating the costs the answer?  If so, this
introduces a level of complexity which does not exist if one pays for
serials from one account.

3) Political problems, etc., etc.

I would be extremely interested in comments from anyone who has made the
transition from a single account for serials to departmental accounts, or
who has some insights into any of the problems which I mentioned.