sketch.* Unfortunately, both on May 10 and May 12, in refitting the gauge, the indicator had to be 

 fixed afresh and therefore the observations cannot be referred to a common zero. 



" A mark was placed on the ship's side to ascertain any vertical movement of the ship relatively to the 

 water surface, and a long plummet was secured in the engine room, to show any alteration in her 

 inclination to the vertical. 



"On April 6, 1903, the tide gauge was re-erected and observations continued, but, owing to the large 

 number of observers employed, the maximum and minimum arrangement was not fitted. 



" The height of the mark (b) on the ship's side above the water was ascertained about once a month in 

 the same manner as during the winter of 1902, i.e., by digging a hole through the ice below the mark and 

 measuring its height. On these occasions the difference between the heights of the leading blocks was 

 measured in the following manner. A wooden scale () marked in half inches was secured to the beam (a) 

 in a vertical position, close to the outer block, with its zero 

 mark on a level with the top of the sheave. A wooden 

 instrument, shaped like the letter T, and having a lead weight 

 attached to its lower end and a hole, A (see fig. 3) in the 

 centre of its upper end, was hung freely on a nail in such a 

 manner that its upper side was horizontal and on a level with 

 the top of the sheave of the inner block. By bringing the eye 

 on a level with the upper side of this T-piece, and noting the 

 position on the scale at which the upper side, if produced, 

 would cut it, the reading of the scale was obtained which gave 

 the difference in height of the blocks. 



"By taking periodical measurements of the height of the 

 mark, and the difference between the heights of the blocks, 

 data were obtained by which readings could be corrected for 

 alteration of the trim and the list of the ship respectively and 

 reduced to a common zero, namely that on April 6, when the tide gauge was erected for the winter of 1903. 



"On September 21, 1903, the wire carried away close to the place where it was secured to the weight 

 resting on the bottom. On examination the wire was found to be greatly 'eaten away, from the point of 

 attachment to a height of several feet, presumably on account of an action between the cast-iron sinkers 

 and the steel pianoforte wire." 



The series of hourly observations was occasionally interrupted by accidents, and the trim and list of the 

 ship changed a little from time to time. Accordingly it is not possible to treat the observations as a 

 continuous whole. The series was therefore broken into a succession of months, so chosen as to avoid 

 periods of manifest irregularity or of accidental interruption, and each month was treated independently. 



The choice of the method of harmonic analysis to be employed seemed to lie between that explained in 

 the "Admiralty Scientific Manual " and that devised for the use of the tidal Abacus.! The method of the 



* No ue has been made of this arrangement. G. H. D. 



t 'Manual of Scientific Inquiry,' Article "Tides"; and "On an Apparatus for Facilitating the Reduction of Tidal 

 Observations," ' Roy. Soc. Proc.,' vol. 52, p. 345. 



I take this opportunity of correcting a mistake in the Manual article, discovered by Mr. SELBT when reducing the " Scotia" 

 tidal observations. At p. 63 



For the tide K ; . In the formula for tan if, in the denominator, for 3 "67 p, read 3 71 p, for a fortnight's observation, 

 and 3 '84 p, for a month's observation. In the formula for H. wherever 3 '67 occurs read 3 71 for a fortnight, and 



3 "84 for a month's observation. The formula H" = H B remains correct. 



3 '67 



For Hie tides K, and P. In the formula for H' the 3 in the numerator (but not that in the denominator) should 

 be replaced by 3-007 for a fortnight's observation, or by 3 '027 for a month's observation. The formula 



Hp = - H' remains correct. 



For K' = K P = (' + V + <f> read ' = n f = f + V + </> + 6 '88 for a fortnight, and K' = K P + f + V + $ + 13" '29 



for a month. 

 Tin' succeeding numerical example must be corrected accordingly. The only sensible i-hnniie is that it' = p = 334 in 



place of 327. 



Fig. 3. 



