TIDE-PEEDICTING MACHINE NO. 2. 31 



Upon the crank shaft, close to the bearing in the desk frame (plate 

 3), is secured a small ratchet wheel. A pawl, kept lifted away from 

 this wheel by friction springs when the machine is moved forward in 

 the right direction, is instantly thrown into engagen^nt when the 

 crank is accidentally turned backward. By pushing in one of the 

 small buttons just above the crank the pawl is locked, which permits 

 of moving the machine backward when desired. Pushing in the 

 other button again secures the machine against accidental backward 

 motion. 



Speed of machine. — The speed with which the machine may be 

 turned, were its object only to produce a tide curve, as is the case 

 with the British machines, varies somewhat with the amplitudes set, 

 a large tide requiring a slower motion of the crank than a small one. 

 The average speed is 20 turns of the crank per minute. A dial day 

 requiring six turns, a year's curve could be traced in 1 hour and 50 

 minutes; but the curve is produced incidentally only to the main 

 purpose of the machine, which is to indicate the times and heights of 

 the high and low waters for filling out the blank forms to be sent to 

 the printer. As ascertained from the predictions of all the compli- 

 cated tides for the years 1912 and 1913, made in the year 1910, the 

 time of setting the machine for a station varies, according to the 

 number of components involved, between 2 and 3 hours for one man; 

 that of operating the machine and copying upon the forms the indi- 

 cated times and heights for one year, from 7 to 12 hours. The dif- 

 ference in the time of the latter operation is due to the greater or less 

 frequency of periods of diurnal or evanescent tides, which require 

 more or less judgment on the part of the operator and therefore cause 

 more or less delay. 



Temperature error. — ^As regards the question of error due to the 

 effects of changes in tempera.ture, which would appear of moment in 

 view of the great length of chain, it may be said that the results are 

 practically unaffected by thermal expansion or contraction. A rough 

 determination of the effect upon the free end of the chain, which in- 

 volves the thermal coefficients of brass and steel and the doubling 

 effect of the motions of the chain pulleys, shows a shortening by 0.055 

 inch or 1.4 millimeters for a rise and a lengthening of the same amount 

 for a fall of 25° F. or 14° C. Such changes of temperature can take 

 place only from season to season, during which time the machine has 

 been readjusted a number of times. 



A possible source of error in the results obtained with the machine, 

 which had, however, been anticipated by the treatment of the chain 

 before described, is that due to wear in its many joints (27.6 feet by 

 125 links equals 3,447) and consequent lengthening of the whole. For 

 the purpose of testing its constancy in this respect, all the components 

 were carefully set to zero and the position of the mean sea level 



