Start with Google Books, which has more metadata and more search options than archive.org. Despite much-publicised metadata failures on some books, most books have useful date, author and title information, and many can be searched inside. I’ve found that the useful search constraints are date (for example 1850 to 1900 when searching for Victoriana) and then one or more of author, title, exact phrase or subject keyword.
Always use the advanced search interface –
Don’t limit the search to “Full Books Only” unless you are searching specifically for epub format ebooks. Epub is a great way of getting a structured copy of a book that you can read on a mobile device, copy and paste from and translate to other formats, but many more books are availabe as PDFs. The only problem with PDF versions of books is that Google Books doesn’t provide downloads for all of them.
Once you’ve found the title of a book you want but that Google doesn’t provide a PDF download for, search for that title on archive.org. It will almost always be there but you may have to select the correct edition, or choose between different versions from different libraries or from projects other than Google Books. Go to the book’s page and don’t click on “Download PDF” which sends you back to Google, who don’t have it. Instead click on the link to “All Files”, then download the PDF (or the DJVU, or the text extracted from the DJVU) from there.
There’s nothing to stop you searching archive.org for authors, title words of keywords as well. archove.org is used by projects other than Google Books. gutenberg.org has the texts of many old books as well, and provides html and epub versions of illustrated books as well. But as a new resource with some rough edges, Google Books is a useful new research resource for historical sources.