SÚ»minaire de thÚ»orie des nombres, Paris 1984-85 Book

Good introduction to the Surreals and overall a good description of the way new mathematical ideas are developed. I like that Knuth decided to stick with Conway's construction of the surreals as cuts between sets, as opposed to other approaches like Gonshor's ordinal-length sequences of +'s and -'s, since this approach makes it easier to build intuition about many important ideas used in later proofs.

The characterization is weak, but that hardly matters. It's a nice attempt to both humanize math and deliver new ideas organically. And I can forgive Knuth a lot of subpar characterization when he does things like make the Surreals appear on a "Conway stone", equal parts Rosetta Stone and Ten Commandments, and the work of the omnipotent deity J. H. W. H. Conway, who created the universe in aleph days, then kept right on creating.