TIDE-PREDICTING MACHINE NO. 2. 33 



The curves traced during the predictions of these hourly heights 

 also served as a means of comparison between results as computed 

 and- given by the machine. The facsimile curve on plate 15, marked 

 Hongkong, 1912, contains the day, December 31, for which the 

 hourly heights are given in Table A. 



In order to ascertain the certainty of action of the machine, pre- 

 dictions were made for a number of days a number of times in suc- 

 cession, the machine being turned back between each set of predic- 

 tions. The agreement between the heights and times of the different 

 sets were within 0.02 foot and 1 minute, respectively. The curves 

 traced in these tests, being on a small scale, revealed no differences 

 when laid upon each other. A practical agreement was also found 

 between the predictions made for the same period by three different 

 observers. 



Amplitude unit. — A tidal calendar giving the tides for the ports 

 and stations of all countries is, like astronomical pubHcations, of 

 international import. It would seem at first glance that the ampli- 

 tude unit of this machine 'might preferably have been based upon 

 the meter rather than the foot, but when it is considered that by far 

 the largest number of charts used by the navigators of all nations 

 give the soundings in fathoms or multiples of the foot, the advantage 

 of using the foot is apparent. Should predictions in meters be 

 required, the conversion of the metric into foot amplitudes, which 

 need only be done once for each station, and the addition of three 

 height-dial rings with scales of 3, 6, and 12 meters, would prepare 

 the machine for the prediction of heights in meters. 



