The advance in knowledge of double and multiple stars has been wonderfully rapid, numerous observers having devoted themselves to this special branch. Many thousands were discovered during the first half of the nineteenth century, and as telescopic power increased new ones continued to flow in by hundreds and thousands, and there has been recently published by the Yerkes Observatory a catalogue of 1290 such stars, discovered between 1871 and 1899 by one observer, Mr. S.W. Burnham. All these have been found by the use of the telescope, but during the last quarter of a century the spectroscope has opened up a new world of double stars of enormous extent and the highest interest. The telescopic binaries which have been observed for a sufficient time to determine their orbits, range from periods of about eleven years as a minimum up to hundreds and even more than a thousand years. But the spectroscope reveals the fact that the many thousands of telescopic binaries form only a very small part of the binary systems in existence. The overwhelming importance of this discovery is, that it carries the times of revolution from the minimum of the telescopic doubles downward in unbroken series through periods of a few years, to those reckoned by months, by days, and even by hours. And with this reduction of period there necessarily follows a corresponding reduction of distance, so that sometimes the two stars must be in contact, and thus the actual birth or origin of a double star has been observed to occur, even though not actually seen. This mode of origin was indeed anticipated by Dr. Lee of Chicago in 1892, and it has been confirmed by observation in the short space of ten years. In a remarkable communication to _Nature_ (September 12th, 1901) Mr. Alexander W. Roberts of Lovedale, South Africa, gives some of the main results of this branch of inquiry. Of course all the variable stars are to be found among the spectroscopic binaries. They consist of that portion of the class in which the plane of the orbit is directed towards us, so that during their revolution one of the pair either wholly or partially eclipses the other. In some of these cases there are irregularities, such as double maxima and minima of unequal lengths, which may be due to triple systems or to other causes not yet explained, but as they all have short periods and always appear as one star in the most powerful telescopes, they form a special division of the spectroscopic binary systems.