FISCHER! TIDE PREDICTING MACHINE 6 



detic Survey from that time until 1910. It has 19 components. 

 It produces no curve, but a record of the results is made by blue- 

 printing the printer's copy. 



Certain weaknesses and considerable wear made it desirable 

 to provide a new machine, which should indicate upon its face, 

 for copying directly upon the printer's form, the exact heights 

 and times of the high and low waters, of the height of the sea at 

 any time, and produce automatically a tide curve with hour and 

 day marks showing the exact time of the high and low waters. 

 Its construction was begun in 1896 and continued with many 

 interruptions, as time could be spared from the regular w r ork, 

 until its completion in February, 1910. 



The principles underlying the machine are the same as those of 

 the English predictors as regards the heights. From the Ferrel 

 machine was adopted the use of two cranks on the same com- 

 ponent shaft: one for summing the cosine terms of the height 

 series and the other, at right angles, for summing the sine terms 

 of the first derivative of the former, the latter, when equal to 

 zero, marking the times of the maxima and minima. 



In order that the machine should be most comprehensive, pro- 

 vision was made for 37 components. The dimensions of its parts 

 were chosen with particular view to rigidity and freedom from 

 flexure. Workmanship and material are such as to reduce 

 deterioration due to wear to a minimum, and the wearing parts 

 are planned so as to permit of replacement or repair without ever 

 interrupting the regular output of the machine. Its dimensions 

 are: length, 10 feet 9 inches; height, 6 feet 2 inches; and width, 

 2 feet. The unit of height amplitude is 0.5 inch. The dial 

 unit of height is 1 inch, 2 inches and 4 inches per foot, according 

 to which of three scales is used. The scale of the record curve, 

 traced on paper 6 inches wide, varies between 1:15 and 1:120, the 

 hour being equal to ^ inch, or the da}' 12 inches. 



To predict a year's tides for a station takes one man from 2\ to 

 4 hours according to the number of components used, for the 

 setting of the machine, and from 8 to 15 hours, according to the 

 complexity of the tides at the station, for copying the indicated 

 results upon the form ready for the printer. 



