OF ARTS AND SCIENCES. 339 



satisfactory results. Evidently the storm front, however determined, 

 must be defined by a feature easily recognized by all observers ; for 

 the usefulness of the relatively few detailed records of the higher class 

 observers depends largely on the accuracy with which they can be 

 placed in their proper position with respect to the storm front, as deter- 

 mined at numerous stations by the simple records of the lower class. 

 It is therefore an important matter to choose what shall be adopted as 

 the guide-observation, as this defines the simplest task that can be set. 



The time of first audible thunder is not satisfactory. It comes 

 without warning:, it is faint; it is not equally perceptible to observers 

 in different situations; even if well observed, it measures a variable 

 distance from the storm-belt, for the flashes that cause the thunder 

 must be of unequal intensity ; finally, it is too distant a product of the 

 storm, and should be replaced by something more immediately con- 

 nected with the line of action. As a guide, it therefore does not seem 

 serviceable. In storms where no rain reaches the ground, the first 

 and loudest thunder-claps are of value ; but such storms are excep- 

 tional, and of small impoi'tance. 



The last thunder is a troublesome thing to note, and it will probably 

 be omitted in the coming summer. The loudest thunder is very useful 

 as a mark of the middle belt, but it is not always accurately definable ; 

 still, it is almost or quite as good a guide as the first rain. But thunder 

 has its chief value in calling the observers to their work ; as soon as it 

 is heard, record it, and then watch for what follows. 



The rain front in well-developed storms is on the whole, I believe, 

 the best guide to their advance across country. The observer is gen- 

 erally on the watch for its arrival, as the thunder in nearly all cases 

 precedes it. It can often be seen coming just before its arrival, and, 

 thus prepared, the observer may note its time within half a minute 

 without difficulty. Even if no thunder is heard before it, the darken- 

 ing approach of the rain clouds is generally sufficient to call one's 

 attention from other occupation. The following records show the 

 measure of agreement that may be expected. They are taken from 

 stations lu northeastern Rhode Island, where observers were close 

 together. 



In the first example, a loud thunder-clap evidently called the atten- 

 tion of all the observers, and in that case gave more closely accordant 

 records than the time of first rain. In the second example, the 

 stations are arranged almost in geographic order, and show an excel- 

 lent sequence of rain times. The agreement is indeed exceptionally 

 good. 



