Micro-crystallite and continuous random network (CRN) models started life i
n the same decade. Their fates have been quite different. With the discover
y of each new family of glasses, micro-crystallite models have been tested
in extreme versions and, rightly, found to be inadequate. CRN models have b
een treated much more elastically, with modifications inserted to accommoda
te new experimental facts. It is now hard to find a clear statement of what
the term CRN means, and most definitions are system-specific. An attempt i
s made here to present a view of the structure of network glasses, which re
cognises the strengths and weaknesses of ordered and random models. and the
impact of recent experimental investigations of medium-range structure. (C
) 2001 Elsevier Science B.V. All rights reserved.