The OAI data feeds don’t explicitly identify volume information in a machine-processable format. (I’m not aware of a standard way of doing this in the formats I’m using, but would be happy to hear suggestions. I’d prefer not to resort to significantly more complicated formats if avoidable.) The Dublin Core version of the metadata, though, includes all the URLs in dc:identifier fields. Any DC record that has multiple identifiers that are all Google can be almost always assumed to refer to successive volumes of a multi-volume work; and the GBS ID is embedded right in the URI.

So you can probably scrape compiled volume data out of my records that way; it’s not the most elegant method, but I believe it will work. I have no corresponding method at present for serial volume ID harvesting short of HTML-scraping their cover pages (though those at least can be identified programmatically in the OAI-PMH feed). I’m happy to entertain suggestions for more elegant methods for harvesting this info for serials and books.

And likewise for serials, to as much of the run as is in Google/Hathi, instead of just the one arbitrary scanned copy I happen to find now.

This is definitely something that’s problematic in the existing GBS and HT interfaces. It’s painstaking to compile this info, but if you’re doing it by hand anyway, please find a way to make it machine-readable!