In addition to the 1997 Edward L Owen column that Rasmus Faber cites, there's another good article here: "Technical origins of 60 Hz as the standard AC frequency in North America" IEEE Power Engineering Review, March 1999, Paul Nixon, p. 35-37. The full article is behind a paywall, but they post the first page as a png image, which I'll link to below.
The particularly interesting section here is:
By late 1889 and early 1890, direct-coupled alternators were coming into the experimental stage. These machines would prove to be much more reliable than the belt-driven generators, but would operate at much lower speeds. The need for lower ac operating frequencies [than 133.3Hz] was apparent, again driven by constructional and mechanical constraints. For example, an alternator direct driven by a 100rpm engine would require 160 poles to yield a frequency of 133 1/3 Hz. This type of construction was viewed as prohibitive. Around this time the Westinghouse Co. conducted an engineering study which considered both electrical operating characteristics with regard to the system components of the time and possible engine driven generator construction constraints, and recommended that 7,200 alternations per minute (60Hz @ 2 poles) was as high a frequency as would be desirable for the engine speeds which were then attainable. 60Hz was actually a carefully selected compromise. It was thought that higher frequencies would be better for the transformers then in existence, while lower frequencies might be better for engine-type generators. 60 Hz first appeared commercially in 1890. The earliest 60 Hz systems, like the earlier ac systems (140, 133 1/3, 125 Hz), were all single-phase.
By 1892, there were a large number of Westinghouse-designed 60 Hz central stations in existence, and 60Hz had taken over a share of the ac business from the higher frequencies.
So we basically have tradeoffs (transformers better at high frequencies vs. electric machines better at low frequencies, which is largely still true today) leading to a somewhat arbitrary compromise, then network effects solidifying the choices.
The Owen article also mentions that southern California was 50Hz up until conversion to 60Hz was completed in 1948. Japan still has half 50Hz and 60Hz: Owen states "In 1895, AEG sold a 50-Hz generator to the power company in Tokyo and the eastern half of Japan was put on the 50-Hz path. A little over a year later, GE sold a 60Hz generator to the power company in Osaka, and the Western half of Japan was put on the 60-Hz path." In the end it seems like it's mostly inertia/network effects -- it's just too painful to change large infrastructure projects.