The Szathmary Recipe Pamphlet Digital Collection is, compared to some of the collections I've waded through, superb. The collection itself was gathered by one Louis Szathmary, a Chicago restauranteer, and a portion of it is now held at the University of Iowa Libraries. We've looked at collections from this library system before, as I recognized parts of the interface.

The objects themselves are "representative samples" from 1880 through 1930. Though it is not explicitly stated as such, the intro blurb on the main page mentions how this era showed a particular shift in eating habits among americans, and that this change is reflected in the ephemera gathered here, so I'd take that as a collection policy. Many characteristics of the objects and collection are covered in the metadata, and with an included "reference URL" link on each object page it appears to be fairly interoperable. One complaint is the zoom function: it is integrated into the object's page, which is nice, but must reload every time you want to move to the left or right, up or down, or zoom in, a tedious operation for someone visually scanning the object up close.

The metadata provided is ample, and easily outshines any collection I've covered thus far. Fields are included not just for the standard Title, Publisher, and Date, but also for all sorts of subject headings (conforming to LCSH, DCMITV and a couple of other acronyms I didn't recognize) and fields including digital collection, contributing collection, and archival collection, as well as Rights Management, Contact, Digitization Specifications and Date Digital. This is a wealth of information, and most terms used in these fields are searchable, i.e. you can click on them and it will bring you all the objects which match that particular criteria. I would say the metadata conforms fully to the guidelines laid out by the NISO framework.

As for the objects themselves, each has an archival quality .TIF image available, and the authenticity can be fairly easily deduced by the substantial metadata. The images are broadly accessible (though they failed the google image search) but are of a very, very large size, as you can plainly see, good for research but not so good for interoperability or use.

I think the target audience of this would be scholars. the stated collection policy mentions what this ephemera reflects, and it's not hard to see a scholar using the robust search capability and zooming ability to easily compare and contrast objects.