Ann,
Thanks for wrestling with the question. I will provide some background, but first let me say that after your reference to Winnie the Pooh, I reminisced a bit with Jefferson Airplane's "House at Pooneil Corners":
Makes you just want to drop acid, doesn't it?
Back to the article count. I have been working on a project to facilitate a negotiation, and have observed that neither party understands the key issues of the other side. For publishers, for example, the essential questions are: How much does it cost us to publish one article (the editorial cost, which is the big one) and, How much do we make for each article we publish? That last question is revenue across all customers combined. For the large commercial publishers that figure is usually in the $5,000-$6,000. For not-for-profit publishers it is usually much higher. Mega-properties (Nature, Science) occupy an entirely different realm. (Note: revenue is not the same thing as profit.) The biggest driver of revenue per article is not price but the reach of global distribution (because more customers means more revenue).
My back-of-the-envelope estimate for the price for the customer per article is around $1.75. That's for the big publishers. It is much higher for not-for-profits. Note that I am not including other revenue streams in this (APCs, advertising, reprints, etc.). It is that $1.75 figure that I am trying to validate (or invalidate).
This metric is different from cost per use (a good measure for librarians, but less meaningful for publishers). I would drop the multiple use category and only include the cost per article for direct sales by the publisher; and the reason for that is simply that that is the way publishers look at their numbers.
As for counting up how many articles a library gets each year, I admit I never realized that this would be a hard number to get to.
Joe Esposito