The Complete Aristotle (eng.)

steps if one wishes to
secure a continuous proof from first principles, or else all
discussion about them wears the air of mere sophistry: for to prove
anything is impossible unless one begins with the appropriate
principles, and connects inference with inference till the last are
reached. Now to define first principles is just what answerers do
not care to do, nor do they pay any attention if the questioner
makes a definition: and yet until it is clear what it is that is
proposed, it is not easy to discuss it. This sort of thing happens
particularly in the case of the first principles: for while the
other propositions are shown through these, these cannot be shown
through anything else: we are obliged to understand every item of
that sort by a definition. The inferences, too, that lie too close
to the first principle are hard to treat in argument: for it is not
possible to bring many arguments in regard to them, because of the
small number of those steps, between the conclusion and the
principle, whereby the succeeding propositions have to be shown.
The hardest, however, of all definitions to treat in argument are
those that employ terms about which, in the first place, it is
uncertain whether they are used in one sense or several, and,
further, whether they are used literally or metaphorically by the
definer. For because of their obscurity, it is impossible to argue
upon such terms; and because of the impossibility of saying whether
this obscurity is due to their being used metaphorically, it is
impossible to refute them.
    In general, it is safe to suppose that, whenever any problem
proves intractable, it either needs definition or else bears either
several senses, or a metaphorical sense, or it is not far removed
from the first principles; or else the reason is that we have yet
to discover in the first place just this-in which of the aforesaid
directions the source of our difficulty lies: when we have made
this clear, then obviously our business must be either to define or
to distinguish, or to supply the intermediate premisses: for it is
through these that the final conclusions are shown.
    It often happens that a difficulty is found in discussing or
arguing a given position because the definition has not been
correctly rendered: e.g. ‘Has one thing one contrary or many?’:
here when the term ‘contraries’ has been properly defined, it is
easy to bring people to see whether it is possible for the same
thing to have several contraries or not: in the same way also with
other terms requiring definition. It appears also in mathematics
that the difficulty in using a figure is sometimes due to a defect
in definition; e.g. in proving that the line which cuts the plane
parallel to one side divides similarly both the line which it cuts
and the area; whereas if the definition be given, the fact asserted
becomes immediately clear: for the areas have the same fraction
subtracted from them as have the sides: and this is the definition
of ‘the same ratio’. The most primary of the elementary principles
are without exception very easy to show, if the definitions
involved, e.g. the nature of a line or of a circle, be laid down;
only the arguments that can be brought in regard to each of them
are not many, because there are not many intermediate steps. If, on
the other hand, the definition of the starting-points be not laid
down, to show them is difficult and may even prove quite
impossible. The case of the significance of verbal expressions is
like that of these mathematical conceptions.
    One may be sure then, whenever a position is hard to discuss,
that one or other of the aforesaid things has happened to it.
Whenever, on the other hand, it is a harder task to argue to the
point claimed, i.e. the premiss, than to the resulting position, a
doubt may arise whether such claims should be admitted or not: for
if a man is going to refuse to admit it and claim that you shall
argue to it as well, he will be giving the signal for a harder
undertaking than was originally proposed: if, on the other hand, he
grants it, he will be giving the original thesis credence on the
strength of what is less credible than itself. If, then, it is
essential not to enhance the difficulty of the problem, he had
better grant it; if, on the other hand, it be essential to reason
through premisses that are better assured, he had better refuse. In
other words, in serious inquiry he ought not to grant it, unless he
be more sure about it than about the