11 



the conditions more perfectly than one of 12 months. There is, theoretically, a minute tidal inequality of 

 long period (LAPI.ACK'S first species) with a period of 14 months due to the variation of latitude, litit it in 

 difficult to see how any perturbation of the lunar semidiurnal tide could lie produced in this way. 



But if we have found a true physical phenomenon, the same kind of effect ought probably to be 

 produced on all the other tides. Yet when the observations for the other tides are plotted out in the same- 

 way, the points appear to be arranged almost chaotically. It is true that some slight tendency may be 

 perceived for an increase of amplitude towards midwinter, but the effect is too uncertain to justify 

 reduction to numbers. 



A much longer series of observations would be needed to throw a clear light on the point raised, but the 

 result is so curious that it would not have been right to pass it by in silence. 



Tidal observations were made at Koss Island (called Erebus Island on the memorandum) by Dr. WILSON 

 from 2 h January 11, 1904, to 8 h January 13. The place of observation was some 40 or 50 miles to 

 the northward of the winter station. As there seemed some reason to suspect a seasonal variability in 

 the tides, it seemed worth while to compare with actuality a tide-curve computed with the constants 

 derived from the winter observations. A curve was therefore run off at the National Physical laboratory 

 for a few days beginning with O 11 January 11, 1904. Although the sites of the two sets of observations 

 are not identical, comparison with actuality shows a satisfactory agreement. It is unfortunate that these 

 observations were made just after the time when the diurnal inequality had vanished and was beginning 

 to increase again; for at these times the agreement is liable to be imperfect between computed and 

 observed curves. On these grounds no surprise need be felt on account of the fact that the semidiurnal 

 tide is somewhat more clearly marked in the observed tide-curve than in the computed one, and that the 

 whole range of the diurnal tide on January 11 was 3 inches greater, and on January 12 about 6 inches 

 (out of 28 inches) greater than appears from the computed curve. The computed and observed times of 

 high and low water agree closely with one another. We may, on the whole, accept these summer 

 observations as proving that our tidal constants are substantially correct. 



The semidiurnal tides, although small, exhibit clearly another peculiarity ; it is that (K of S-.>) - (K of Mj) 

 exhibits a seasonal change of roughly the same character in both years. 



In all cases " the age of the tide" is negative and its mean value is about - 4 days; in other words, 

 spring-tide occurs four days before or ten days after full and change of moon. 



If the phases of M 2 and S;> differed by 180 we should have neaps at full and change, and springs at half 

 moon. This case corresponds to " direct" lunar tide and " inverted" solar tide. In the actual case 



(K of M 2 ) - (K of So) = 370 - 272 = 98 ; 



thus the observations show a result a very little nearer to this condition than to the ordinary one where 

 springs coincide with full and change of moon. 



The unusual relationship between the M 2 and S 2 tides is such as to make it worth while to examine 

 what would be the condition of affairs in an ocean of uniform depth covering the whole planet. From the 

 few soundings which have been made it would seem that the ocean may be about 600 fathoms in depth, 

 although further north the depth appears to be considerably greater. I have therefore taken the formul* 

 of Mr. Hoi-ini (' Phil. Trans.,' A, 191 (1878), pp. 177, 180) and evaluated the lunar and solar semidiurnal 

 tides for an ocean of 7260 ft. in latitudes 60% 65, 70, 75 with the following results: 



Lrx.ui Semidiurnal Tide. 



C 2 



