324 BULLETIN: MUSEUM OF COMPARATIVE ZOOLOGY. 
tematic attempt to discover the facts by which these deductions might 
be confirmed. 
In the case of a tilting, with fulerum at the river mouth and at right 
angles to the general river course, the maximum height of two-swing 
terrace scarps would be found somewhere near the middle course of the 
river; and the scarp height would thence decrease down-valley and up- 
valley. It seems that, in such a terrace system as that of the Connecti- 
ent, it may be possible to apply this test and thus gain from the 
dimensions of the terraces a direct proof as to the kind of movement by 
which the work of terracing was initiated, as well as a confirmation 
of the evidence already in hand regarding the nature of postglacial 
movement in the New England province. 
RevaTioN OF THE PreceDING DEDUCTIONS TO THE OBSERVATIONS 
DESCRIBED IN THE Fonnowina Sections. The facts presented in the 
following sections are chiefly details of structure and form, directly 
observable ; changes of form are occasionally noted, but these are of 
relatively small measure. All these details are but the present members 
of a long series of facts, every one of which might have been recorded, 
had observers been living to witness them; and then the origin of ter- 
races would be fully understood. But the earlier members of the series 
are hopelessly lost to observation from being prehistoric. In their 
unavoidable absence, theory attempts to supply a series of conditions, 
pictured by the reasonably guided imagination, which shall imitate the 
series of past facts, and thus, as it were, call them to life, bring them’ 
into the field of vision. The success of the theory is not to be measured 
so much by the apparent reasonableness of its fundamental suppositions, 
or by the definiteness with which various imaginary consequences may 
be deduced from it, as by the accuracy with which the observable 
members of the deduced consequences imitate the facts of actual occur- 
rence. The greater the number of peculiar categories of observed facts, 
the greater the probable correctness of a theory whose deduced conse- 
quences can match all of them. Hence the importance of minute 
observation and careful generalization on the one hand, and of accurate 
and detailed deduction on the other. Hence also the importance of 
carefully distinguishing these unlike processes in order that their results 
may be systematically confronted in an unprejudiced comparison. The 
elaboration of the deductions in the preceding sections therefore seems 
to be as necessary a part of the study of terraces as is the accumulation 
of observations for presentation in the following sections. 
The theory of terracing has here been presented before the observa- 
