Writing after Parmenides, Anaxagoras saw the importance of his predecessor’s conception. In a month Anaxagoras could verify Parmenides’ insight: the perceived shape of the moon is a function of its position relative to the sun, showing that the moon is indeed illuminated by the sun. Parmenides’ insight allowed Anaxagoras to turn the problem of the moon’s phases into a geometrical exercise, and to solve the problem.
He could also infer that the moon was a spherical body, since the shape of the shadows could only appear on a sphere; that the moon was opaque and reflective, and had to be orbiting below the sun or, in other words, nearer to the earth than the sun; if not, it would not become invisible (completely in shadow) at the time of the new moon. Furthermore, his theory established that the sun shone all night, because in the middle of the lunar month (around the time of the full moon) the moon was visible all night long, reflecting the sun’s light. The sun, therefore, was not extinguished at night, as Heraclitus thought, nor did it continue traveling onward to the west, as Xenophanes thought, nor was it hiding behind some mountains to the north, as Anaximenes thought. No, it was below the earth, still shining, circling eastward, all the night long, heading for its rising the next morning.
Most important, at the end of one lunar month and at the beginning of another, when the moon disappeared from view, the orb still had to exist, a dark, opaque body, lurking in the vicinity of the sun (from the viewer’s perspective) but (as we have seen) below it. The moon was, as one could see at certain times when both bodies were visible, approximately the same apparent size as the sun. Thus there was a possibility that the moon might, as it moved below the sun, pass in front of the sun and, since the moon was opaque and thus solid, block the sun’s light.
The solar eclipse on February 17th allowed Anaxagoras the opportunity to test his hypothesis. The eclipse happened at the new moon, the only time it could happen according to his hypothesis. The darkened portion of the sun was circular, and just slightly smaller than the disk of the sun. Finally, by determining the limits of the umbra (the shadow of the moon), Anaxagoras verified the eclipse had occurred by the moon’s screening the sun’s light — by antiphraxis as Aristotle would later call it.[14]
Anaxagoras’ model also explained how lunar eclipses could happen. When the sun and the moon were in opposition, that is, 180º from each other, the earth, which was much larger than the other bodies (as he thought), would lie between them. If the sun and moon happened to line up with the earth right between them, then the earth would block the sun’s rays to the moon. This could happen only in the middle of the month, when the moon was full. And that was exactly right. Anaxagoras had taken Parmenides’ insight, verified its truth, developed its implications, applied the theory to new problems, and solved the mystery of eclipses at one stroke.
But if Anaxagoras was right about eclipses — and he was — the theories of all his philosophical predecessors were wrong, hopelessly wrong. All early theories saw the moon as generating its own light. They also saw the heavenly bodies as inherently light structures of gas or fire. For Anaxagoras, by contrast, the heavenly bodies were massy, heavy bodies composed of earth or stone. They were not fed by vapors, but illuminated by friction, or perhaps by being in a region of aether, or luminous gas. Since the bodies were heavy, they could not remain aloft by virtue of their lightness, or by being blown in the wind. A cosmic whirl, the dinē or vortex, kept the bodies in motion by a kind of centrifugal force. Anaxagoras’ contemporary Empedocles said the motion was like that of a twirled ladle in which the liquid did not fall out. (Presumably ancient bartenders used to spin ladles of wine to impress their guests.) But if heavenly bodies were kept in motion by a cosmic turbine, a little jolt could send one of them hurtling toward earth.
[14].Aristotle Posterior Analytics 90a15-18.