_Variable Stars._—It has often happened in the history of different branches of physical science that observation and experiment were so far ahead of theory that hopeless confusion appeared to reign; and then one chance result has given a clue, and from that time all differences and difficulties in the previous researches have stood forth as natural consequences, explaining one another in a rational sequence. So we find parallax, proper motion, double stars, binary systems, variable stars, and new stars all bound together. The logical and necessary explanation given of the cause of ordinary spectroscopic binaries, and of irregular proper motions of Sirius and Procyon, leads to the inference that if ever the plane of such a binary orbit were edge-on to us there ought to be an eclipse of the luminous partner whenever the non-luminous one is interposed between us. This should give rise either to intermittence in the star’s light or else to variability. It was by supposing the existence of a dark companion to Algol that its discoverer, Goodricke of York,[16] in 1783, explained variable stars of this type. Algol (β Persei) completes the period of variable brightness in 68.8 hours. It loses three-fifths of its light, and regains it in twelve hours. In 1889 Vogel,[17] with the Potsdam spectrograph, actually found that the luminous star is receding before each eclipse, and approaching us after each eclipse; thus entirely supporting Goodricke’s opinion. There are many variables of the Algol type, and information is steadily accumulating. But all variable stars do not suffer the sudden variations of Algol. There are many types, and the explanations of others have not proved so easy. The Harvard College photographs have disclosed the very great prevalence of variability, and this is certainly one of the lines in which modern discovery must progress. Roberts, in South Africa, has done splendid work on the periods of variables of the Algol type.