A longstanding puzzle concerns the calculation of the gluino condensate [tr
lambda(2)/16 pi(2)] = c Lambda(3) in N = 1 supersymmetric SU(N) gauge theo
ry: so-called weak-coupling instanton (WCI) calculations give c = 1, wherea
s strong-coupling instanton (SCI) calculations give, instead, c = By examin
ing correlators of this condensate in arbitrary multi-instanton sectors, we
cast serious doubt on the SCI calculation of [tr lambda(2)/16 pi(2)] by sh
owing that an essential step-namely cluster decomposition-is invalid. We al
so show that the addition of a so-called Kovner-Shifman vacuum (in which [t
r lambda(2)/16 pi(2)] = 0) cannot straightforwardly resolve this mismatch.
(C) 2000 Elsevier Science B.V. All rights reserved.