CLYDE SEA AREA. 689 



fall abnormals, displaced forward by six weeks, from the curve of actual density. The 

 result was that the June maxima were each split into a pair of maxima separated by a 

 small minimum, while in other parts of the curve the amplitude alone is changed and not 

 the form; the subsidiary maxima of December and May are reduced, but all the minima are 

 lowered to an even greater extent. The curve of bottom salinity having less amplitude, 

 and being much more uniform, the scale of the curve of deviations subtracted from it was 

 only one-tenth of that used in the previous case. This curve, uncorrected, has a distinct 

 relation to the rainfall ; maxima occur in August and September, and a single well- 

 marked minimum in February. The subsidiary maximum of the surface curve in 

 December is indicated by a flattening of the descending curve. The result of applyiDg 

 the correction is to produce the effect of a double maximum in July and September, 

 separated by a slight minimum in August, and to bring out a slight minimum in 

 December with a slight maximum in December, but it is otherwise unchanged. 



The rainfall, although generally under the average amount, was as a rule higher and 

 •lower month about, so that a two-monthly curve, giving the year's fall by a line drawn 

 through six (instead of twelve) points, would be almost parallel to the normal curve, only 

 the maxima occurred in September and November instead of October and December. 

 The annual variation of rainfall may be said to account fairly for the annual variation of 

 salinity observed. But, according to the rainfall, there was no occasion for the very 

 low surface minimum of February 1887, and the maxima of December 1886 and May 

 1887. The bottom salinity, in fact, bears a closer relation to the rainfall than does the 

 surface salinity. In February the low salinity is fully explained by melting snow, and by 

 the hard frost which prevailed during the trip freezing the freshened surface water which 

 was headed up Loch Fyne and Loch Strivan by winds blowing up the loch, while the 

 other lochs were calm or traversed by light cross breezes. In December Upper Loch 

 Fyne was calm, ice-laden, and very fresh, the comparatively high salinity of the landward 

 portion was contributed by Lochs Goil, Long, and the Gareloch, down which a strong 

 wind was blowing, raising deep salt water at the loch heads. There is no apparent 

 reason for the subsidiary maximum of May 1887. The subsidiary minimum of June 

 1887 is obviously a wind effect, as in that month there was a steady breeze up the lochs 

 tending to head back the fresher water, and allow whatever circulation took place to be 

 at the expense of the deeper layers, in contrast to June 1886 when a general down-loch 

 wind prevailed. 



The conclusion to be drawn from this consideration is, that the calm condition of the 

 weather, postulated as an essential for the critical comparison of the rainfall and salinity 

 curves, did not prevail, and that consequently it is impossible to deduce the normal 

 salinity curve with any degree of confidence. But an approach to this can be made by an 

 arbitrary assumption, smoothing the curves of actual salinity and rainfall by taking the 

 two-monthly means of the latter, and allowing for wind and frost effects in the former. 



The curve of normal surface salinity for the landward division, as traced in this way, 

 shows one yearly maximum in August three months after the minimum rainfall of May, 



VOL. XXXVI. PART III. (NO. 23). 5 L 



