MERIDIAN ASTROMICTRY — BOSS. 205 



such motions and they can be classified to some extent as to origin, if not as 

 to their full meaning. As a rule the total amount of such systematic motions 

 amounts to only a few seconds during a century ; but in the investigations 

 by which motions for each star are ascertained, and through which organ- 

 ized motion may be recognized, it is of the first importance to get rid of the 

 systematic errors of observations which tend to neutralize or exaggerate our 

 estimate of the motions that actually exist. We must attempt, so far as we 

 can, to free the measurements of star-positions from their sj'stematic errors. 



This task has occupied close attention here for several years past. The 

 result has been full of encouragement. Various tests, some of them unfore- 

 seen, indicate that the adopted corrections are probably fair approximations 

 to fact. 



It should be understood that more or less work of this nature is going on 

 iiere all the time, and that perfecting the tables of systematic correction for 

 some of the star-catalogs of greater extent (though often of minor accuracy) 

 has absorbed the efforts of a considerable percentage of our computing force 

 during the past 5^ear. 



In the Preliminary General Catalog we now have effective means for 

 extending computations of systematic correction to minor star-catalogs and 

 lists that contain few observations of so-called standard stars. We shall 

 thus add to our resources much valuable material not hitherto available in 

 our deduction of proper motion. Much of this dates back forty years or 

 more, and will be specially useful. 



At the same time, we carefully distinguish between systematic corrections 

 and the effects of errors of observation that are really nothing else than 

 fortuitous accumulations of error. Each case must be considered on its 

 own merits, and must be subjected to adequate tests and checks sufficient 

 to indicate the real nature of differences between the standard and the 

 catalog of observation. 



For several good reasons our discussion must also apply to as many stars 

 as practicable. The discussion of systematic motions of the stars implies 

 selection and discrimination as to objects. For instance, if we wish to 

 determine whether the constants of solar motion are the same or different 

 for different spectral types of stars, the total number of available stars must 

 be divided into different classes, and the value of our conclusions will largely 

 depend upon sufficiency of numbers of stars in each class. We shall prob- 

 ably find, for instance, that the few stars of the Orion type, as well as those 

 of Class IV, indicate a low angular velocity of solar motion, as if those stars 

 are either at a relatively great distance or as if their general drift is approxi- 

 mately parallel to the motion of the sun. At any rate, we shall find it 

 necessary to deal with all available examples of these two classes of stars, 

 and it will appear that even the entire number maj^ be insufficient for a 

 thoroughly satisfactory conclusion in respect to them. 



