If the markings were evidently of a permanent nature and attached to the solid shell of the planet, and if they were of sufficient distinctness to be seen in substantially the same form by all observers armed with competent instruments, then whatever conclusion was drawn from their apparent motion as to the period of the planet's rotation would have to be accepted. In the case of Mercury the markings, while not easily seen, appear to be sufficiently distinct to afford confidence in the result of observations based upon them, but Venus's markings have been represented in so many different ways that it seems advisable to await more light before accepting any extraordinary, and in itself improbable, conclusion based upon them. It should also be added that in 1900 spectroscopic observations by Belopolski at Pulkova gave evidence that Venus really rotates rapidly on her axis, in a period probably approximating to the twenty-four hours of the earth's rotation, thus corroborating the older conclusions. Belopolski's observation, it may be remarked, was based upon what is known as the Doppler principle, which is employed in measuring the motion of stars in the line of sight, and in other cases of rapidly moving sources of light. According to this principle, when a source of light, either original or reflected, is approaching the observer, the characteristic lines in its spectrum are shifted toward the blue end, and when it is retreating from the observer the lines are shifted toward the red end. Now, in the case of a planet rotating rapidly on its axis, it is clear that if the observer is situated in, or nearly in, the plane of the planet's equator, one edge of its disk will be approaching his eye while the opposite edge is retreating, and the lines in the spectrum of a beam of light from the advancing edge will be shifted toward the blue, while those in the spectrum of the light coming from the retreating edge will be shifted toward the red. And, by carefully noting the amount of the shifting, the velocity of the planet's rotation can be computed. This is what was done by Belopolski in the case of Venus, with the result above noted.