one of those highly civilised architects who flourished a thousand years
earlier or eight hundred later. But such exceptions are rare. As a rule
primitive art is good--and here again my hypothesis is helpful--for, as
a rule, it is also free from descriptive qualities. In primitive art you
will find no accurate representation; you will find only significant
form. Yet no other art moves us so profoundly. Whether we consider
Sumerian sculpture or pre-dynastic Egyptian art, or archaic Greek, or
the Wei and T'ang masterpieces,[1] or those early Japanese works of
which I had the luck to see a few superb examples (especially two
wooden Bodhisattvas) at the Shepherd's Bush Exhibition in 1910, or
whether, coming nearer home, we consider the primitive Byzantine art of
the sixth century and its primitive developments amongst the Western
barbarians, or, turning far afield, we consider that mysterious and
majestic art that flourished in Central and South America before the
coming of the white men, in every case we observe three common
characteristics--absence of representation, absence of technical
swagger, sublimely impressive form. Nor is it hard to discover the
connection between these three. Formal significance loses itself in
preoccupation with exact representation and ostentatious cunning.[2]

Naturally, it is said that if there is little representation and less
saltimbancery in primitive art, that is because the primitives were
unable to catch a likeness or cut intellectual capers. The contention is
beside the point. There is truth in it, no doubt, though, were I a
critic whose reputation depended on a power of impressing the public
with a semblance of knowledge, I should be more cautious about urging it
than such people generally are. For to suppose that the Byzantine
masters wanted skill, or could not have created an illusion had they
wished to do so, seems to imply ignorance of the amazingly dexterous
realism of the notoriously bad works of that age. Very often, I fear,