116 EFFECTS OF WINDS AND OF 



29, were inspected in detail on the graphs. This inspection showed evidence 

 in 16 cases out of the 25 that a short-period oscillation (3.7 hours) and a 

 long-period oscillation (13.0 hours) were apparently both in progress at the 

 same time. In five other cases out of the 25 the inspection showed evidence 

 that an oscillation with a period of about 3.7 hours was apparently in progress 

 complicated by other things, but no 13-hour period was evident. 



At this point in the investigation it appeared to be probable that the 13- 

 hour seiche was almost or quite continuously present at Buffalo, usually 

 more or less masked by other seiches and by forced oscillations. Accord- 

 ingly, the principal maxima and minima were selected from among the 

 196 tabulated in table No. 29 to cover the whole 22 days and to correspond 

 to the supposition of the preceding sentence. From these selected principal 

 maxima a second determination of the long period became available, and, 

 similarly, from the selected principal minima a third determination. 



The three determinations of the period of the long Buffalo seiche as thus 



determined were : 



Period 



equals 



From frequency-distribution table, No. 29, mean of 10 cases 13.0 hrs. 



From principal maxima throughout the 22 days, 28 cases 13.4 hrs. 



From principal minima throughout the 22 days, 31 cases 12. 8 hrs. 



Mean 13.1 hrs. 



The mean 13.1 hours was adopted as the most probable value of the 

 period of the long seiche, and its uncertainty was estimated at not much 

 more than 0.1 hour. 



Again, the graphs were carefully studied in detail for evidence as to 

 apparent damping of the 13.1 hour seiche and for evidence as to the impulses 

 which initiate it. Three conclusions were reached : 



(1) Each new large 13.1 hour seiche is preceded by a large discernible 

 impulse from the wind or from barometric pressure. (2) Large 13.1 hour 

 seiches are subject to rapid apparent damping even though some new impulse 

 occurred. (3) A wind impulse, if sufficient to change the elevation of the 

 water surface corresponding to equilibrium by more than 0.20 foot, ordina- 

 rily either started a new 13.1 hour seiche or clearly distorted such a seiche 

 which was already in progress. 



SEICHES AT CLEVELAND. 



At Cleveland 22 days of hourly elevations of the water surface observed 

 at the gage were studied in detail for seiches. These particular 22 days had 

 been especially selected to include periods of unusually large barometric 

 effects and wind effects, especially the latter. It was to be presumed, there- 

 fore, that these days included the impulses which started many new seiches 

 of unusual size. 



The study at Cleveland was somewhat similar to that indicated above as 

 having been made at Buffalo. The frequency-distribution table for Cleve- 



