PhiUp Lake — Bala Lake and Ricer Sysfem-i 209 



to the right ; while in the direct line there is a slight depression, 

 which is certainly filled with drift, however deep or shallow that 

 drift may be. 



Here, again, it is impossible to assert that the solid floor of the 

 valley is higher than the bottom of the lake. And as no more 

 rock is seen in silu until we pass below the level of the lowest point 

 of the lake, we are compelled to admit that in this direction there 

 is no certain evidence that the lake is rock-bound. 



So far, then, as this evidence goes, Bala Lake may have been 

 formed in the same way as the lakes of Cumberland and West- 

 moreland. The sections show that, like them, it is a submerged 

 river valley. Its general direction and the slope of its bed indicate 

 that the valley of which it is a portion is the valley of the Bala 

 fault. Originally, no doubt, the waters drained south-westward into 

 the sea at Barmouth (or, perhaps, at Towyn). But the channel 

 was subsequently blocked, and the drainage of the upper portion 

 was forced to seek a fresh outlet. 



I shall, however, give reasons for supposing that, although the 

 old channel may have been choked by drift, the reversal of the 

 drainage was due, at least in part, to earth-movements ; and that 

 it was these earth-movements which determined the position of the 

 watershed. 



Relation of the Faults to the Valley and to the Watershed. 



That the valley to which, as we have seen, Bala Lake belongs, 

 is in the line of a gi'eat fault, has long been known ; and the fault 

 is represented on the Survey maps as passing down the middle 

 of the lake. It lies, I believe, more nearly along the south-eastern 

 shore, and the opposite shore coincides with the line of a second fault, 

 which, like the formei', has its downthrow towards the lake. Thus 

 the lake lies in the continuation of a trough, and its two sides 

 coincide or nearly coincide with the two faults enclosing the trough. 



South of Pant-gwyn the valley is very narrow. Its north-western 

 side is formed of volcanic rocks, which are usually supposed to be of 

 Arenig age, while the south-eastern side is formed of Sliddle Lingula 

 beds containing Lingulella Davisi. But the floor of the valley is 

 concealed by drift and alluvium, and there is nothing to indicate 

 the exact position of the fault which undoubtedly exists, nor is 

 there any evidence to show whether this fault is simple or compound. 



At Pant-gwyn, however, the valley opens out ; the faults are 

 much more clearly seen, and here they form a complex system 

 which we shall now proceed to examine. (Fig. 3.) 



It is, perhaps, unnecessary to give in detail the evidence on which 

 these faults are drawn. It will almost be sufficient to say that, 

 in general, wherever the volcanic series exists, it is conspicuously 

 visible, and it is an easy matter to trace its boundary. 



The slates below the volcanic series are also well enough exposed, 

 but it is diflerent with the post-volcanic slates in the valley ; and 

 as many of the exposures of these beds are difficult to find, it may 

 be well to enumerate them. 



DECADE IV. — VOL. Til. — NO. V. 14 



