in South- Western Victoria. 235> 



The shell beds, both at Roscoe's and the Devil's Den, are 

 always near the main banks of the river, and when found 

 along the one, they are generally absent from the other. I 

 have seen no shells towards the middle of the flats, but 

 these are, for the most part, covered by alluvium and sandy 

 drifts, which conceal the underlying strata. Every flood, 

 however^ leaves behind a fresh deposit of shells in the bed, 

 and on the margin of the stream, washed out from the 

 fossiliferous portions of the terrace by storm waters. I am 

 told that when a landslip of the black alluvium takes place, 

 the shells are exposed in great numbers. Beneath the 

 superficial accumulations there are probably numerous beds 

 of shells, not perhaps regularly stratified, but forming 

 banks in various parts. 



The terrace just described is not, however, the only one 

 in this part of the river's course, as between it and the 

 water's edge there is usually another and much narrower 

 one. This lies about 20 feet below the upper terrace, and 

 extends for 20 or 30 yards only on either side of the river. 

 I could find but few shells in it, and these few were always 

 near the surface, as if they had been simply washed there 

 from the higher ground. 



The spots referred to as so thickly strewn with shells lie 

 at a lower level still, and are merely the sloping banks of 

 the river channel itself The shells deposited upon them 

 can, of course, only be gathered when the water is low. 



The main banks of the river are at a height of from 130 to 

 160feet above its bed, and thus from 70 to 100 feet higher than 

 the upper terrace, the ascent from which to the banks above 

 is often steep, and sometimes even perpendicular. The 

 terraces follow closely the windings of the river, but the 

 topmost banks bound its general course only. The Glenelg 

 flows for about two-thirds of its length through a very wide, 

 as well as deep, valley, though the river itself is but a 

 moderate one. The depth of the gorge is tolerably uniform, 

 but its width varies considerably, being greatest in the 

 neighbourhood of the fossil beds, where it is, as I have said, 

 nearly two miles across. It is about half this width a few 

 miles to the north, and narrows also, though more gradually, 

 to the south of Limestone Creek. 



With the summit of the Glenelg banks commence the- 

 wide plains forming the heath and scrub lands of Follett 

 and South Normanby. Occasionally, there is a gentle rise of 

 20 or 30 feet before the tableland is reached, though, as a 



