20 JJniversity of California Publications in Zoology [Vol.15 



3. The distance the boat moves while the observations are being 

 taken. 



In a quiet sea the pelorus "ive.s fairly accurate results but, when 

 rough, the error may be very large. For this reason it is seldom used, 

 except for cheeking against sextant observations, unless atmospheric 

 conditions make the use of the .sextant impossible. Positions deter- 

 mined by it may be considered accurate within 0.3 to 0.5 miles. 



Dead-reckoning is used only wlien all other methods are imprac- 

 ticable; namely, during fog, or at night, or when the distance from 

 land is so great as to render points of bearing invisible. The method 

 consists in measuring the speed of the boat along its course by means 

 of a log-line. It involves large errors resulting from the following 

 causes : 



1. Difficulty in accurately determining the departure point, such 

 as the extreme edge of a kelp bed. 



2. Difificult.v of holding the boat on the desired course, which varies 

 ^\^th the "sea" as well as the .skill of the pilot. 



3. Drift and leeway due to currents, direction of wind and sea, 

 and velocity of wind and sea, all of which alter the coiirse of the boat 

 and cause the log-line to over or under-read. 



4. Deviation in accuracy of the compass, depending upon number 

 and position of anchors and other magnetic materials. 



The latter error is reduced to a minimum by "swinging the ship" 

 and making our corrections for each heading, but the effect of the 

 other errors is such that their magnitude increases with the distance 

 covered by the boat. Care is taken, however, to "cheek back" after 

 reaching port or anchorage in order to correct the courses and distances 

 registered in the captain's log-book. With the utmost care, however, 

 an error of one mile in position sometimes occurs. 



Finally, each of the above methods is ased as a check against the 

 others. In addition soundings when made are always compared with 

 the entries on the charts and the position thus determined is cheeked 

 against that derived by the three methods above described. In this way 

 the average error in latitude and longitude, relative especially to our 

 work since 1908, was reduced to 0.3 miles, although in some instances 

 the error is upwards of one mile (see p. 45). 



(&) Depth.— All our determinations of depth depend upon a meter- 

 wheel which registers the amount of cable paid out. It is obvious, 

 therefore, that unless the cable be perpendicular, the registration will 

 exceed the depth. In order to measure the error due to such inclina- 



