14 UNITED STATES COAST AND GEODETIC SURVEY. 



heights at any time, thus avoiding the long process of measuring a 

 curve, suggests itself at the first study of the subject. 



Time dials with their pointers in fixed relations to the driving 

 shafts of the components were already used in the Ferrel machine. 



A simpler and more direct use of the means already employed in 

 the Ferrel machine for pointing out the maxima and minima of the 

 height function was found in the coincidence of a marked link in the 

 moving chain which sums the first derivative of the height series, 

 with a fixed index in view of the operator. At the instant when this 

 marked link is at the index the sum of all the derivative heights is 

 zero and, consequently, the height either a maximum or a minimum. 

 When the marked link approaches the index from the right it is a 

 high water and when it approaches it from the left it is a low water. 



In order to have at the office a record of the predictions copied 

 from the face of the Ferrel machine directly upon the forms to be 

 sent to the printer, careful copies were made of them by hand for 

 some years, after which blue prints or photostat copies of the original 

 tabulations were used. Tide curves, as produced by the British 

 machines, serving as such records, suggested the idea of saving the 

 time and expense of making copies by having the free end of the 

 height summing chain, in addition to pointing out the heights on 

 the dial, also move a pen upon a moving strip of paper and produce a 

 tide-curve record automatically.* The marking of the latter into 

 time spaces is a process employed for many years in various ways in 

 the chronograph, meteorological recording instruments, etc., and 

 was accomplished in the British tide predictor No. 3 by means of 

 the momentary lateral displacement of the curve tracing pen. 



Before making the first plans the subject of mechanical summation 

 was exhaustively studied with the hope of improving upon the chain 

 and pulley, but all the methods suggesting themselves, when devel- 

 oped in detail, led back to that, the simplest and most efficient form 

 of summing harmonic series. A number of practical tests were also 

 made for the purpose of studying the relative merits of bevel and 

 spur gears. The former style of gear wheels having been replaced 

 in the third British machine by the latter to obtain greater rigidity, 

 and the Ferrel machine showing some flexure when set up with 

 large amplitudes, made caution as to this point necessary; espe- 

 cially as the machine was to provide for so large a number of com- 

 ponents. Besides, the shape and dimensions of the machine as a 

 whole depended largely upon the decision as to this point. Having 

 decided in favor of the use of bevel gears and upon the dimensions 

 of the detail parts necessary to insure rigidity, and a half-inch unit 

 for the height amplitudes having been generally accepted as most 

 suitable, a drawing was prepared, upon the scale of 1:10, which was 

 submitted to and accepted by the instrument board sometime in 1895, 

 and the work of construction was begun in 1896. 



