But, side by side with this inquiry, another had been in progress. A crowd of diligent labourers had been searching with close and rigid scrutiny into the circumstances attending ancient eclipses. A new light had been thrown upon this subject by the labours of modern travellers and historians. One remarkable instance of this may be cited. Mr. Layard has identified the site of Larissa with the modern Nimroud. Now, Xenophon relates that when Larissa was besieged by the Persians, an eclipse of the sun took place, so remarkable in its effects (and therefore undoubtedly total), that the Median defenders of the town threw down their arms, and the city was accordingly captured. And Hansen has shown that a certain estimate of the moon’s motion makes the eclipse which occurred on August 15, 310 B.C., not only _total_, but _central_ at Nimroud. Some other remarkable eclipses—as the celebrated sunset eclipse (total) at Rome, 399 B.C.; the eclipse which enveloped the fleet of Agathocles as he escaped from Syracuse; the famous eclipse of Thales, which interrupted a battle between the Medes and Lydians; and even the partial eclipse which (possibly) caused the ‘going back of the shadow upon the dial of Ahaz’—have all been accounted for satisfactorily by Hansen’s estimate of the moon’s motion: so also have nineteen lunar eclipses recorded in the Almagest. This estimate of Hansen’s, which accounts so satisfactorily for solar and lunar eclipses, makes the moon’s rate of motion increase more than twice as fast as it should do according to the calculations of Adams. But before our readers run away with the notion that astronomers have here gone quite astray, it will be well to present, in a simple manner, the extreme minuteness of the discrepancy about which all the coil has been made. Suppose that, just in front of our moon, a false moon exactly equal to ours in size and appearance (see _note_ at the end of this paper) were to set off with a motion corresponding to the present motion of the moon, save only in one respect—namely, that the false moon’s motion should not be subject to the change we are considering, termed _the acceleration_. Then one hundred years would elapse before our moon would fairly begin to show in advance. She would, in that time, have brought only one one-hundred-and-fiftieth part of her breadth from behind the false moon. At the end of another century she would have gained four times as much; at the end of a third, nine times as much: and so on. She would not fairly have cleared her own breadth in less than twelve hundred years. But the _whole_ of this gain, minute as it is, is not left unaccounted for by our modern astronomical theories. _Half_ the gain is explained, the other half remains to be interpreted; in other words, _the moon travels further by about half her own breadth in twelve centuries than she should do according to the lunar theory_.