22 SMITHSONIAN MISCELLANEOUS COLLECTIONS VOL. 66 



for this work is too uncertain. The most careful deduction of the 

 velocity is certainly very important in order to at least make known 

 the degree of probability that the corresponding planetary orbit 

 departs to one side or the other from the parabola. The computation 

 of the elements, if we really wish to attempt it, should be carried out 

 with various assumed and appropriate values of the velocities, 

 whereby the general result can be better arranged for comparison 

 with other cases. 



III. RESULTS OF OBSERVATIONS AND COMPUTATIONS 

 RELATIVE TO METEORS 



I. Average Accuracy of Observations and Computations 



A. AT THE END OF THE ORBIT OR POINT OF ARREST 



The average error of a given direction (of an azimuth) as the 

 result of 351 observations is found to be ±5.8 degrees. 



These determinations were made in 12 per cent of the cases by 

 simultaneous reference to stars, and in about 20 per cent of the cases 

 on the basis of subsequent measurements, in the remaining cases by 

 reference to terrestrial objects in the neighborhood by means of plans 

 and charts. 



The average error of any indication or description of the apparent 

 altitude or zenith distance may be taken as ±4.1 degrees from 235 

 cases. 



As a general rule in these cases we have considered only direct 

 references to stars or subsequent measurements. Crude estimates, 

 as is well known, give almost always apparent altitudes that are far 

 too large. When such cases have been used in exceptional cases, we 

 have in general reduced them to f or f of the given estimates. 



The azimuths were not examined so closely since in such cases 

 errors in one direction are less to be feared, which is also easily to be 

 seen by reason of the larger average error of a single observation. 



The average error in the determination of the geographic position 

 of the terminal point as deduced from 42 cases amounts to ±8.3 

 kilometers. The very best determinations are uncertain by 3^ or 4 

 kilometers. The average error of the computed linear altitudes 

 amounted to ±3.4 kilometers. 



B. AT THE INITIAL POINT OF THE OBSERVED PATH 



The nature of the observations brings it about that a meteor can be 

 first observed by many observers at very different points of its path. 

 These are differences that do not possess any of the characteristics 



