On RYTHMICAL MEASURES. 6r 



fuccefhon as uniformly as he can, without greatly regarding 

 the numbers. When we attempt the divifion into five, we are 

 often difpofed to break down the interval into two parts *, and 

 fubdivide the one part into three, and the other into two, ma- 

 king the two either equal to each other, or in the proportion of 

 one and two. In the fame manner, when we attempt the divi- 

 fion of an interval into feven, we are perhaps irrefiftibly led to 

 halve it, giving three to the one half and four to the other. A 

 divifion into any of the higher primes is impracticable. 



We have not only the power of dividing and fubdividing 

 fmall intervals or units of time in the manner I have explained, 

 but we can alfo form aggregates of them, by telling off equal 

 parcels of them, when they are rendered fenfible to the ear by 

 repeated ftrokes or founds. We do this, by attending particu- 

 larly to the firft of each of the parcels. The confequence of 

 this is, that thopgh the fucceffive flrokes give out exactly the 

 fame found, and are made with the fame degree of force, we 

 fhall infallibly imagine the firft of each of our parcels to be 

 fomewhat louder or more ftrongly expreffed than the others. 

 When we have counted off one parcel, the mind confiders the 

 whole as one adl ; the parcel is as it were fet by, and a new 

 effort is made to count off the next parcel. Hence the firft of 

 this parcel, being marked by a particular effort of the mind, 

 and as it were reprefenting the whole parcel, ftrikes it more 

 ftrongly than the reft, and \s therefore conceived as louder ; and 

 the laft of the parcel, being leaft attended to by the mind, is 

 conceived as feeble, thus giving us the impreflion of fomething 

 like a break or paufe betwixt the parcels. Thus, when we liften 

 to the beats of a watch, we are at firft difpofed to reckon them 

 by pairs ; and we invariably find, that the firft of each pair is 

 confidered by us as a ftrong and the other as a feeble found. 



We 



* More efpecially if, in former parts of the fucceflion, the fame interval has been 

 found frequently divided into two. 



