One of the less talked about examples of copyright misuse is what’s known as copyfraud. This is where a person or organization asserts copyright (or copyright-like) restrictions on a work to which they have no legal rights to do so. This is most commonly perpetrated by archival institutions such as libraries, historical societies, and museums attempting to restrict the access and use of materials in their collections to which they do not own the copyright or where the copyright has expired.
For simplicity’s sake let’s use the example of an historic 19th century photo that is in the public domain in the United States, and it’s in the collection of a local library. That library has every right to restrict access to said photo by requiring an in-person visit to view it, by providing only a low-resolution copy on their website, or even by watermarking it. If someone wants to use that photo in a publication, and get a better copy that’s print-worthy, then the library can charge money for the high-resolution, non-watermarked version, whether digitally or with photographic reproduction. They’re essentially charging for access to the original. That’s generally fine.
Where things start to get iffy is when they attempt to restrict any further use of the image, even the low-resolution watermarked copy on their website. Since the image is in the public domain, they have no right to impose any restrictions on the further use of that image, or at least they shouldn’t. But here’s where the questions about watermarks and terms of use come into play.
The 1999 Bridgeman Art Library v. Corel Corp. case in the United States District Court for the Southern District of New York determined that slavish reproduction of a public domain work does not confer copyright on the reproduction since it is not transformative nor creative (even if highly skillful). Essentially, format-shifting a public domain work does not grant a copyright to the new format, so libraries cannot claim copyright on their scans. While it wasn’t a Supreme Court case and is not binding precedent, it is highly influential as persuasive authority, and it is widely followed by other federal courts.
I have not been able to find a definitive answer to whether adding a watermark is sufficient to transform a scan from a “slavish reproduction” to a “transformative/derivative work”. I would argue that it isn’t, but like I said, I haven’t been able to find any information dealing with it. Most information out there is just about whether you need a watermark to show that you own the copyright to an image you produced yourself (you don’t) or if it really proves any sort of ownership (it doesn’t). But putting your own watermark on a scan you made of a public domain image is starting to get into questionable territory.
Another thing that some libraries do, especially on their websites, is have a click-through agreement or other terms of use that you are required to accept in order to access their archives. This can include the same kinds of conditions that you find in copyright law, but they’re not actually asserting copyright, just these new conditions. Is that able to trump fair use? Of course they can say anything they want in their agreement, but that doesn’t make it enforceable.