As a matter of fact, he has such a key at the telescope which he uses to make his observations in taking time, so that when he wishes to record the precise instant in which anything takes place which he is viewing through his telescope he has but to press the key in his hand and an extra tooth will be put into the record which the clock is making, somewhere among the regular teeth put in by the beating of the clock. Later, when he takes out the sheet he can see just where the tooth came, and so at what instant it was. If, now, he knows exactly what the instant was according to the true time as given in his almanacs--that is, what his clock ought to have shown at that instant--he can tell how nearly right his clock is. Once knowing how this clock is, it is a simple calculation to find how the clock which sends the signals is running, and to alter it if needed in a manner we shall describe later. The observations the astronomer makes use of to determine these instants of time are upon the "clock stars." He uses a rather small telescope, known as a transit. It is placed with the nicest accuracy in a north-and-south line. It can turn over vertically, but can not move sideways out of its line. Its alignment is kept perfect by occasionally sighting some small mark a few rods from the observatory, either north or south. If the astronomer points this transit, say, halfway up the southern heavens and sees a star pass across the center of its field he knows that that instant gives, as it were, the "noon mark" of that star. If it is one of the "clock stars," he knows by his tables what that instant of time is--should be--by his clock. We have seen what his means are of comparing his clock and his observations. But observe, now, how much pains he takes to get the most exact observations. To begin with, he must have calculated to a nicety his location. The director of an observatory always knows where he is located in a sense that few other men do. The accuracy of a large part of his observations of any kind depends on his first having determined the latitude and longitude of his observatory within a very few feet. Then the data given by his tables are all modified, and adapted to conform to his locality.