If you publish your work in the United States, the reproductions of the Rosenwald sheet in books like Kaplan's Encyclopedia
would not be considered to be under copyright. This precedent was set by the case Bridgeman Art Library v. Corel Corp
. in 1999, and has not been challenged since.
http://en.wikipedia.org/wiki/Bridgeman_ ... Corel_Corp
This is the policy under which wikimedia commons operates.
http://commons.wikimedia.org/wiki/Publi ... lic_domain
- scroll down to "Interaction of US copyright law and non-US copyright law"
Since the Rosenwald sheet is in the public domain, and it is physically also in a US institution (the National Gallery of Art in Washington), then "faithful reproductions" of it are not copyrightable, and can be used for any purpose, including commercial. I assume this would also mean that if you (re)published it anywhere in the world, it would not violate any copyright, since both the object and its reproductions come from the US.
This issue hasn't been tested in any European court, but most institutions and publishers in Europe claim ownership of photographic reproductions of two-dimensional images in their collections (as well as rights over photographs of three-dimensional objects, so that, for instance, you are not allowed to publish a photo of the pyramid at the Louvre that you took unless you first ask for permission of the Louvre). Whether or not these claims would stand up in court I don't know, but I am not afraid to use their images for non-commericial purposes without asking for permission. If they want to find me and ask me to stop, let them do so.