but separate, and following each other. Hoek, of Utrecht, showed some years ago that such comet-families exist. When we compare the orbits of the comets of 1843 and 1668 there is nothing that forbids the idea of their identity. The differences are no greater than probable perturbations might account for. Then, again, the comets of
Fig. 4.
1843 and 1880 may easily be identical. Indeed, the orbit given for the latter comet corresponds to a period of almost thirty-seven years, and Meyer has shown that the observations can not be reconciled with a period less than thirty or greater than fifty years. Now, thirty-seven years would take us back just to 1843, so that it is very likely that these two comets are really one and the same. So far the "identifiers" have matters their own way. But, now, as to the comet of 1882. Can it be identical with the comet of 1880? We think not. The orbit of the latter was computed exclusively from observations taken after its perihelion passage, so that no action of the sun depending upon its close approach at perihelion can account for its return in less than three years, and the inclination of its orbit is such that ever since it went out of sight it has been out of harm's way as to perturbations by the planets. Then, again, the orbit of the comet of 1882 does not agree with the idea of identity. Whatever other effects may have been produced by the resistance of the solar atmosphere at perihelion, this resistance must have tended to shorten its period, if it changed it at all. Now, the observations thus far taken, though per-
