February 7, 1919] 



SCIENCE 



137 



than the errors of observation. Attempts to 

 represent these deviations by means of empir- 

 ical formula? have so far failed to meet the 

 test of prediction, and it begins to look as if 

 they were, in the strict sense of the word, 

 irreg-ular^though of the fundamental jieriod- 

 icit>' lying behind them there can be no doubt. 



The maximum brightness, and the details 

 of the form of the light curve, also differ very 

 considerably at different times for the same 

 star. The mean of a number of periods, how- 

 ever, usually gives a fairly smooth light-curve 

 of roughly sinusoidal form, but with the rise 

 usually steeper than the fall. Harmonic anal- 

 yses of these curves have been made by Turner 

 and Phillips, showing that the first harmonic 

 largely predominates though the second and 

 third are usually quite sensible. Phillips has 

 shown that nearly all the stars so far investi- 

 gated fall into one or other of two groups, 

 marked by certain definite relations between 

 the phases of the second and third harmonics. 

 In other words the deviations of the light 

 curve from a simple sinusoid tend to follow 

 one or other of two definite patterns. This 

 discovery will doubtless prove of theoretical 

 importance, but no attempt to explain it can 

 yet be made. 



The above remarks apply to the curves ob- 

 tained by plotting the stellar magnitude 

 against the time. If the actual light-emission 

 should be used instead, the cun'es would have 

 high, steep-pointed maxima and very flat 

 minima, and their representation by a Fourier 

 series would demand a host of harmonics. 

 The simple character of the curves giving the 

 magnitude as a function of the time suggests, 

 as Plummer has pointed out, that the periodic 

 process involved ma.v be something, such as va- 

 riation in the thickness of an absorbing layer, 

 which would directly affect the logarithm of 

 the escaping light. But such a layer would 

 have to absorb over 90 per cent, of the energy 

 passing through it for months at a time with- 

 out becoming greatly overheated, which seems, 

 hard to believe. 



The most notable spectroscopic features of 

 these stars are the presence of heavy flutings 

 of titanium oxide, indicating very low surface 



temperatures, and of bright hydrogen lines — 

 which are especially strong near maximum; 

 a:id, on the negative side, a conspicuous ab- 

 sence of changes in radial velocity. 



The rather scanty data at present available 

 indicate that the peculiar velocities of these 

 stars are ver.v high, and their proper motions 

 moderate or small in amount. This would 

 indicate considerable distance and luminosity, 

 and it seems clear that, at maximum, these are 

 fairly bright giant stars, at the least. Further 

 data, especially regarding proper motions, are 

 much to be desired, and should be obtainable 

 in many cases. 



Very little is known of the real causes of 

 variation of this type. It is certainly not due 

 to orbital motion, and the prevalent irregular- 

 ities suggest strongly that we have here to do, 

 not with a gravitational or rotational phe- 

 nomenon, but with a physical process — some- 

 thing resembling in nature the eruptions of a 

 geyser, when an acciunulation of internal 

 energj- piles up to the iwint where il obtains 

 relief through some overlying resistance, 

 giving rise to roughly periodic outbursts of 

 varying intensity. The relativel.y great length 

 of the period falls in well with this h.ypothesis. 

 There is much about the phenomena which 

 suggests the variation in solar activity of 

 which the sunspots are the most prominent 

 symptoms. This is prett.y certainly due to 

 some such accumulation of internal strain 

 and has the same irregularly periodic char- 

 acter, though a much longer period. 



What tlie actual nature of the process is 

 which can change the brightness of a star by 

 several hundred fold remains, however, for the 

 future to determine. 



Mention should be made in passing of two 

 curious t.'i-pes of variation, each represented by 

 but few stars, but very definite, which are 

 classified at Harvard as subdivisions of class 

 II. The typical star of one of these groups. 

 R Corona? Borealis is usually of about the 

 sixth magnitude, but at irregular intervals 

 drops rajiidly to the eighth, tenth, or even the 

 thirteenth magnitude, recovering again after 

 intervals which may be months or years in 

 duration. Two or three other stars behave in 



