Aha, thinking to google search for "proquest z3950" actually got me some 
additional clues!

"Sites that are currently using Z39.50 to search ProQuest are advised to 
consider moving to the XML gateway."

in Google snippets for:

http://www.proquest.com/assets/downloads/products/techrequirements_np.pdf

Also "If you are using the previous XML
gateway for access other than with a federated search vendor, please 
contact our support center at
www.proquest.com/go/migrate and we can get you the new XML gateway 
implementation documentation."

Okay, so now I at least know that something called the "XML Gateway" 
exists, and that's what I want info on or ask about!  (Why are our 
vendors so reluctant to put info on their services online?)

I am not a huge fan of z3950, and am not ordinarily optimistic about 
it's ability to actually do what I need, but I'd use it if it was all 
that was available; in this case, it seems like Proquest is recommending 
you do NOT use it, but use this mysterious 'XML gateway'.



On 2/12/14 3:29 PM, Eric Lease Morgan wrote:
> On Feb 12, 2014, at 3:22 PM, Jonathan Rochkind <[log in to unmask]> wrote:
>
>> I feel like at some point I heard there was a search API for the
>> Proquest content/database platform.
>
>
> While it may not be the coolest, I’d be willing to bet Proquest supports Z39.50. I used it lately to do some interesting queries against the New York Times Historical Newspapers Database (index). [1] Okay. I know. Z39.50 and their Reverse Polish Notation query language. Yuck. Moreover, the bibliographic data is probably downloadable at MARC records, but hey.
>
> [1] Z39.50 hack - http://blogs.nd.edu/emorgan/2013/11/fun/
>
> —
> Eric Lease Morgan
>
>