"It has been remarked respecting variable stars, that in passing through their successive phases, they are subject to sensible irregularities, which have not hitherto been reduced to fixed laws. In general they do not always attain the same maximum brightness, their fluctuations being in some cases very considerable. Thus, according to Argelander, the variable star in _Corona Borealis_, which Pigott discovered in 1795, exhibits on some occasions such feeble changes of brightness, that it is almost impossible to distinguish the maxima from the minima by the naked eye; but after it has completed several of its cycles in this manner, its fluctuations all at once become so considerable, that in some instances it totally disappears. It has been found, moreover, that the light of variable stars does not increase and diminish symmetrically on each side of the maximum, nor are the successive intervals between the maxima exactly equal to each other."--_Grant's History of Physical Astronomy_, p. 541.
Of the numerous hypotheses hitherto proposed to account for these phenomena we believe none can be found to include and harmonize all the facts of observation. The theories of Herschel and Maupertius fail to explain the irregularity in some of the periods; while those of Newton and Dunn afford no explanation of the periodicity itself.[27] But let us suppose that among the fixed stars some have atmospheres of great extent, as was probably the case with the sun at a remote epoch in its history. Let us also suppose the existence of nebulous rings, like those of our own system, moving in orbits so elliptical that in their perihelia they pass through the atmospheric envelopes of the central stars. Such meteoric rings of varying density, like those revolving about the sun, would evidently produce the phenomena of variable stars. The resisting medium through which they pass in perihelion must gradually contract their orbits, or, in other words, diminish the intervals between consecutive maxima. Such a shortening of the period is now well established in the case of _Algol_. Again, if a ring be influenced by perturbation the period will be variable, like that of _Mira Ceti_. A change, moreover, in the perihelion distance will account for the occasional increase or diminution of the apparent magnitude at the different maxima of the same star. But how are we to account for the variations of brightness observed in a number of stars where no order or periodicity in the variation has as yet been discovered? It is easy to perceive that either a single nebulous ring with more than one _hiatus_, or several rings about the same star, may produce phenomena of the character described. Finally, if the matter of an elliptic ring should accumulate in a single mass, so as to occupy a comparatively small arc, its passage through perihelion might produce the phenomenon of a so-called temporary star.