Shown oy Recording Tide-gauges. 133 



occasions on which the period shifts from 4 - 7 m to 12*9 m, 

 would also mean some marked change in the currents through 

 Cabot Strait. 



Does the varying depth of the Bay of Fundy affect the period at 



St. John f 



As stated in an earlier paper (already referred to), a conclu- 

 sive test of the seiche theory would be the following. The 

 period of oscillation of a body of water varies inversely as the 

 square root of the mean depth. Now the mean depth of the 

 Bay of Fundy at low water is something over 200 ft., and at 

 high water (on the average) nearly 20 ft. greater. The width 

 of the bay at St. John being about 40 miles, it can be calculated 

 that the period of secondary undulation at St. John should, 

 according to the seiche theory, be about 1*8 m. greater at low 

 water than at high water. 



To test this I selected 106 tracings of particularly well- 

 marked secondary undulations at St. John. These I measured 

 for period with the greatest care. Then 1 sorted them out 

 into high tide undulations and low tide undulations, finding 49 

 of the former and 57 of the latter. The mean period for 

 high tide was 40*5 m. and for low tide 41*7 — a difference of 

 1*2 m. Now this was wholly confirmatory of the seiche theory. 

 That the difference 1*2 should come out less than the cal- 

 culated difference 1*8 is readily understood from the fact that 

 the undulations nearly always continued before and after high 

 and low tide, so that the effective difference of depth would 

 be less than the mean range of the tide. 



To test the probability of the above differences being mere 

 chance, I calculated " the probable error " of each series, and 

 found it to be the same — 0*4 m. Now the mean of each 

 series only differs from the mean of the whole by 0*6 m. 

 From this it is a comparatively simple problem to calculate 

 what is the probability that by mere chance the mean 

 period at high tide was 0*6 m. higher than its real value and 

 the mean value fur low tide the same amount lower. The 

 probability is y 1 -^ or the chances are 12 to 1 against the differ- 

 ence being due to the mere chance distribution of somewhat 

 rough determinations of what is really the same quantity. 



Thus this test, while very strongly in favor of the seiche 

 theory, is not conclusive. 



Connections between undulations at different places. 



Taking first points on the Atlantic and the adjacent Bay of 

 Fundy, it is shown by a comparison of the records at Halifax, 

 Yarmouth and St. John, that marked oscillations at one are 



