seventh month was to be secured before the observations ceased for the season, but the choice of the stage 

 at which the overlapping should be made to occur was more or less arbitrary. 



Seventh month. This begins with O h November 13, 1902. The observation for 22 h of December 9 is 

 missing and was supplied by interpolation. On December 1 3 the wire parted and the series ended for the year. 



In the second winter, that of 1903, somewhat greater care seems to have been taken to note the small 

 shifting of the ship. 



Eighth month. This begins with O h April 6, 1903. Between April 22 and May 3 the ship shifted so aa 

 to make the readings too high by 3 inches, compared with the earlier ones. As an arbitrary correction I 

 deducted 1 inch from all heights from O h April 24 to O h April 27 ; from l h April 27 to O 1 ' April 30 I 

 deducted 2 inches ; and for the rest of the month the full 3 inches. These arbitrary corrections were 

 submitted to independent harmonic analysis, and it appeared that they afforded corrections so minute aa 

 to leave the tidal constants virtually unchanged. 



Ninth month. This begins with O h May 9, 1903. The ship shifted considerably at some time about 

 June 12, and as it is only possible to obtain one month before that date, there is an unutilized gap of a 

 few days between this month and the one before it. 



Tenth month. This begins with O h June 15, 1903. On July 10 a sensible shift in the trim and height 

 of the ship was discovered. This necessitates the addition of 4 inches to all heights, 2J inches being due 

 to angular movement and 2 inches to vertical movement. As an arbitrary correction I added 2 inches to 

 all heights from O h July 8 to O h July 9 ; and afterwards I added the full 4| inches. 



Eleventh month. This begins with O h July 14, 1903. 



Twelfth month. This begins with O h August 14, 1903. 



After September 8 the observations were only taken every two hours, and for the remainder of the 

 month the values at the odd hours were interpolated. 



The observations stop on September 20, but are not used in the reductions after September 13. 



No corrections have been applied for changes in the barometric pressure. As the application of such a 

 correction would have been very laborious and, moreover, somewhat speculative, I have relied on the 

 automatic elimination of the inequalities produced by taking mean values. 



The following are the results of the twelve harmonic analyses, the heights being slated in inches : 



N.B. The values of A represent merely the changes in the position of the ship and have therefore no 

 physical significance; all the heights arc stated in inches. 



The values of H and K are somewhat irregular from month to month, and it is therefore not permissible 

 to adopt the mean values of H and K as representing the mean tide. I have therefore formed H cos K and 



