From: William Gunn <[log in to unmask]>
Date:  Mon, 26 Jan 2015 13:43:40 -0800

OK, so I got curious about this and decided to take Kathleen's
analysis one step further. I got the data from Ulrich's for all active
peer-reviewed scholarly journals that are available online. I got
12561 total journals (smaller than Kathleen's number probably due to
the "scholarly" restriction).

Of those, only 4798 list prices in either USD, GBP or EUR (and a
handful in other currencies). The median price of these titles is
$701, with the lowest being $29.10 and the highest at $23701. The
distribution is very narrow, however, with most journals listing in
the $200-250 range, as you see can see
here:https://www.dropbox.com/s/viyzo0hnqh914tu/frequency%20distribution%20of%20price.png?dl=0

Just in case you're curious, the most expensive listed journals are (in USD):

$23707 - "Journal of Polymer Science. Part A, Polymer Chemistry"
$23201 - "Molecular Crystals and Liquid Crystals"
$21638 -  "Ferroelectrics"
$16682 - "Journal of Neuroscience Research"
$15164 - "Advanced Materials"

The sum of all the listed journals is $5,615,592, and since they
represent about 40% of the total, we can extrapolate to get
$15,494,542.

So Kathleen's estimate was pretty close given the data on hand, but
note the absence of Nature, Cell, Science, etc from this list. Most of
the journals which do list prices seem to be technical trade journals
- the kinds of journals that might sell a non-insignificant number of
subscriptions to industry. So this might have little relation to what
an academic library might end up with were they to seek to pay a la
carte for each subscription, or it might be a fairly accurate
reflection of what would happen if the market was 100% a la carte with
no bundling.

Just speculating for a moment, academic libraries generally pay less
than industry, so a la carte prices would probably also be lower than
the (I guess) industry prices shown here. Looking over the list prices
for some journals that I do know the academic a la carte price for, I
don't think including them in the list would skew the distribution all
that much, making the above extrapolation not entirely unreasonable.

William Gunn
+1 (650) 614-1749
http://synthesis.williamgunn.org/about/