I beg to differ. In any 3 phase system, if two hots are used to feed a motor or other device and a third supply conductor as a neutral (whether or not present) is not incorporated into the circuitry, then the power received by the device is also theoretically a pure (think: balanced) sine wave although at the voltage (e.g. 208) thought of as "odd" by someone just getting acquanted with 3 phase power. (When the AC peak is reached on one leg while the valley has not yet been reached on the other leg due to the 120 degree phasing relationship, the result is still a sine wave. The "other" leg will eventally reach valleys and peaks of the same magnitude as those reached by the first leg.)
Given an ancient Edison direct current power station providing plus 100, minus 100, and neutral (zero referenced), I find it hard to believe that equipment and especially light bulbs and motors were designed back then (1900) to allow indiscriminate use on either 100 volts (hot to neutral) or 200 volts (hot to hot). The design of devices including refrigerators would be somewhat complicated and not worth the expense.
Instead it would be much more reasonable to provide each customer (each home) with hot and neutral at 100 volts as the service. The practice of supplying neutral and one hot persisted proliferously well into the alternating current era in the U.S., with what is still referred to and in some cities still available as two wire 120 volt service.