hands had been subjected. Thus, we see that a language of 'senses' is not a very reliable language, and that we cannot depend on it for general purposes of evaluation.
How about the term 'dog' ? The number of individuals with which any one is directly acquainted is, by necessity, limited, and usually is small. Let us imagine that someone had dealt only with good-natured 'dogs', and had never been bitten by any of them. Next he sees some animal ; he says, 'This is a dog'; his associations (relations) do not suggest a bite; he approaches the animal and begins to play with him, and is bitten. Was the statement 'this is a dog' a safe statement? Obviously not. He approached the animal with semantic expectations and evaluation of his verbal definition, but was bitten by the non-verbal, un-speakable objective level, which has different characteristics.
Judging by present standards, knowledge in the days of Aristotle was very meagre. It was comparatively easy 2300 years ago to summarize the few facts known, and so to build generalizations which would cover those few facts.
If we attempt to build a-system, 1933, can we escape the difficulties which beset Aristotle ? The answer is that some difficulties are avoidable, but that some are inherent in the structure of human knowledge, and so cannot be entirely evaded. We can, however, invent new methods by which the harmful semantic effect of these limitations can be successfully eliminated.
There is no escape from the fact that we must start with undefined terms which express silent, structural creeds or metaphysics. If we state our undefined terms explicitly, we, at least, make our metaphysics conscious and public, and so we facilitate criticism, co-operation,. The modern undefined scientific terms, such as 'order', for instance, underlie the exact sciences and our wider world-outlook. We must start with these undefined terms as well as the modern structural world-outlook as given by science, 1933. That settles the important semantic point of our structural metaphysics. It need hardly be emphasized that in a human class of life, where creeds are characterized by having dates, they should always be labelled with this date. For sanity, the creeds utilized in 1933 should be of the issue of 1933.
Now as to the structure of our language. What structure shall we give to our language ? Shall we keep the old structure, with all its primitive implications and corresponding s.r, or shall we deliberately build a language of new structure which will carry new modern implications and s.r? There seems to be only one reasonable choice. For a-system, we must build a new language. We must abandon the 'is' of identity, to