these occasions evidently have a higher temperature, and this is not unnatural, since the sun is losing heat by radiation from its outer portions. We know that the temperature of our atmosphere decreases with greater heights. The movements of the air are concerned in this change. A sinking mass of air is compressed by the increased pressure to which it is being exposed, and its temperature rises, therefore, just as the temperature rises in a pneumatic gas-lighter when the piston is pressed down. If the air were dry and in strong vertical motion, its temperature would change by 10° Cent. (18° F.) per km. If it stood still, it would assume an almost uniform temperature; that is to say, there would be no lowering of the temperature as we proceed upward. The actual value lies between the two extremes. As the gravitation in the photosphere of the sun is 27.4 times greater than on the surface of the earth, we can deduce that, if the air on the sun were as dense as on the earth, the temperature on the sun would vary 27.4 times as much as on the earth with the increasing height—that is to say, by 270 degrees per kilometre, provided its atmosphere were in violent agitation. Now, the outer portions of the solar atmosphere are, indeed, in violent motion, so that this latter assumption seems to be justified. But this part consists essentially of hydrogen, which is 29 times lighter than the air. We must, therefore, reduce the value at which we arrive to one-twenty-ninth. As a result, the final temperature gradient per kilometre would only be 9° Cent. (16.2° F.). But the radiation is extremely powerful on the sun, and it tends to equalize the conditions. Nine degrees per kilometre is therefore, without doubt, too high a value. Further, in the interior of the sun the gases are much heavier. At a small depth, however, they will be so strongly compressed by the upper strata that their further compressibility will be limited, and the calculation which we have just made loses its validity. Yet, in any case, the temperature of the sun must increase as we penetrate nearer to its centre. If we accept a temperature gradient per kilometre of the value above indicated, 9°—it is three times greater in the solid earth-crust—we should obtain for the centre of the sun a temperature of more than six million degrees.