But careful observation showed that there were marked irregularities in the light fluctuations, indicating that Eros either had a very strange distribution of light and dark areas covering its surface, or that instead of being a globular body it was of some extremely irregular shape, so that as it rotated it presented successively larger and smaller reflecting surfaces toward the sun and the earth. One interesting suggestion was that the little planet is in reality double, the two components revolving around their common center of gravity, like a close binary star, and mutually eclipsing one another. But this theory seems hardly competent to explain the very great fluctuation in light, and a better one, probably, is that suggested by Prof. E.C. Pickering, that Eros is shaped something like a dumb-bell. We can picture such a mass, in imagination, tumbling end over end in its orbit so as to present at one moment the broad sides of both bells, together with their connecting neck, toward the sun, and, at the same time, toward the observer on the earth, and, at another moment, only the end of one of the bells, the other bell and the neck being concealed in shadow. In this way the successive gain and loss of sixfold in the amount of light might be accounted for. Owing to the great distance the real form of the asteroid is imperceptible even with powerful telescopes, but the effect of a change in the amount of reflecting surface presented produces, necessarily, an alternate waxing and waning of the light. As far as the fluctuations are concerned, they might also be explained by supposing that the shape of the asteroid is that of a flat disk, rotating about one of its larger diameters so as to present, alternately, its edge and its broadside to the sun. And, perhaps, in order completely to account for all the observed eccentricities of the light of Eros, the irregularity of form may have to be supplemented by certain assumptions as to the varying reflective capacity of different parts of the misshapen mass. The invaluable Harvard photographs show that long before Eros was recognized as an asteroid its light variations had been automatically registered on the plates. Some of the plates, Prof. E.C. Pickering says, had had an exposure of an hour or more, and, owing to its motion, Eros had formed a trail on each of these plates, which in some cases showed distinct variations in brightness. Differences in the amount of variation at different times will largely depend upon the position of the earth with respect to the axis of rotation.