some others, and is perhaps true for all. These bright lines, clue to incandescent hydrogen, undergo various modifications during the time in which they are present. The relative intensity of different lines varies greatly in different stars, and also in the same star at different phases. Mrs. Fleming has been able to arrange them in a series having ten subdivisions, with R Lyncis at one end, with the H β and H γ lines prominent and H δ wanting, and R Leonis at the other end with H β wanting, H γ faint, and H δ prominent. There are also corresponding changes in the distribution of the remaining light of the spectrum, a peculiarity which is shared by stars of the same type which are not variable. These characteristics are well shown in Fig. 7.
Their great range of variation makes many of these stars invisible when near minimum in telescopes of ordinary size. This may account for the custom which has been followed by many observers of measuring the light only when the star is near maximum. This is unfortunate, since the determination of the length of the period is not sufficient in itself for the solution of the problems involved. On this account special efforts have been made at the Harvard Observatory, where the observations are carried on by Miss A. J. Cannon and Mr. Leon Campbell, to get measures of the variables at all the different phases. Even thus it is doubtful if the secrets of the changes can be found, until the research is made to include a more detailed spectroscopic study than has yet been made. A systematic study of a large number of wellselected stars is much needed. This could probably be done best by a photographic reflector of the largest size. Such a scheme of work has been proposed by Professor W. W. Campbell, director of the Lick Observatory, and from it we may expect results of the highest value. It may be well, also, to study the radial motions of these variables, but it is more than doubtful if their variability is in any way associated with orbital motion, such as would be found in binary systems. The irregularity in the recurrence of the phenomena seems to preclude the possibility of such an explanation. The stars of this class probably contain within themselves the causes of their changes. They are, perhaps, at that critical stage of development where occasional internal disturbances cause tremendous outbursts, especially of incandescent hydrogen, resulting in an enormous increase of light. The commotion slowly dies down only to return again with more or less of regularity. For the details of these disturbances we must await further study.
IV. Of the 3,000 variables known at present probably the vast majority have short periods, that is, periods of a few days, or a few hours. The periods, also, are uniform; or, at least, if apparent irregularity exists at times, this is capable of being expressed by rigorous mathematical formulæ, η Aquilæ and β Lyræ are well-known examples of this class. Recent investigations have shown that such stars are binary systems, and that in some way the light phenomena are as-
