426 D. MACKINTOSH ON THE ERRATIC BLOCKS OF 
the derivation of which I traced many years ago. They must 
have been carried by a sheet of land-ice fifteen miles in breadth, or 
by great masses of floating ice; and the blocks must have fallen 
where the air or the water was sufficiently warm to melt the ice so 
as to prevent their further transportation; for they terminate too 
abruptly, and on too extensive a scale, to admit of any other suppo- 
sition. If land-ice, according to the principles of dynamical science, 
could not have moved 170 miles on level or undulating ground, 
without the slightest general fall in the direction of the movement, 
but, on the contrary, a very considerable rise towards the extremity 
of its course, then floating ice must have been the transporting agent 
(see II. § 8). Numerous other instances, showing the importance 
of boulders as a means of determining the kind of ice-action, will be 
described in the sequel. Boulders are likewise important as a means 
of correlating the different deposits of clay, gravel, or sand in which 
they are imbedded. In this respect they are often more reliable 
than marine shells, because the latter may have varied with the “ sub- 
marine climate” of the area over which the invariable erratics were 
dispersed. Marine shells, however, are very important as indica- 
tions of changes of climate during the accumulation of drift-deposits 
in vertical succession, as lately shown by Mr. Shone, F.G.S., of 
Chester, and Dr. Gwyn Jeffreys, F.R.S. (Quart. Journ. Geol. Sec. for 
May 1878). 
2. Generalization of Facts not premature.—lt has been suggested 
that for some time to come observers of glacial phenomena should 
content themselves with collecting facts, and leave the task of gene- 
ralization to some one who at an indefinitely remote period may be 
able to collect the results of numerous observations, and frame a 
general theory of the glacial episode. It ought, however, to be con- 
sidered that no authors can generalize facts so well as those who 
haye discovered or observed them, and that the progress of science 
may be obstructed, instead of advanced, by philosophers making 
systems out of phenomena with which they are not practically 
acquainted. A mere theorizer is apt to be influenced by what may 
be called @ prior likelihood ; whereas a practical observer of glacial 
phenomena finds himself often reduced to the necessity of admitting 
what he would never have expected. This arises partly from the 
abnormal character of the phenomena with which he has to grapple. 
3. Identification of Boulders—Tests and Oountertests—To find out 
the parent rock of a boulder, of course requires that specimens should 
be compared; but one specimen of each is seldom sufficient, and, 
unless the rock be very peculiar in composition and structure, it is 
not always safe to rely on a number of specimens. Where the cha- 
racter of the parent rock is varied and it is possible to trace a 
similar variety in a single boulder or group of boulders, the work of 
identification becomes less hazardous. In the case of Criffel granite 
there is a variety in the rocks i situ which is represented among 
the boulders scattered over the plain on the opposite side of the 
Solway Frith, and likewise prevalent on the beach north of Park- 
gate, Cheshire. One variety strongly resembles Mull granite (which 
