HathiTrust is much better than Google Books about allowing access to works that are no longer under copyright in the United States. Under US law, everything published 1929 and before is currently in the public domain. But there are a lot of special cases where 20th century works published after 1929 are also in the public domain:

https://guides.library.cornell.edu/copyright/publicdomain

Google Books appears to follow the blanket 1929 rule, or did the last time I looked. HathiTrust has cleared the copyright status for many additional works following the more complex rules, e.g.

"Drawing Birds" by Joy Postle, 1953:

https://babel.hathitrust.org/cgi/pt?id=nyp.33433115876140&se...

Unfortunately, the Google-originated scans that HathiTrust has come with special restrictions. Google itself required that only people associated with the academic libraries could download whole books as a unit, even for works that are in the public domain:

https://hathitrust.atlassian.net/servicedesk/customer/portal...

Fortunately, members of the public can download individual page scans without any special affiliation. People have naturally written tools to automate this process so that full books can be reassembled and then uploaded to the Internet Archive or other book sites.

Google Books has a much faster and sometimes better search interface, so a common flow I use is to search Google Books for terms and then go to HathiTrust to read inside books that Google Books surfaced but won't show.

EDIT: corrected 1926 to 1929 per cxr's comment below.