To determine factors governing triplet repeat expansion at FMR1, we ne
ed to understand the basis of normal variation. We have sequenced the
FMR1 repeat from 102 normal X chromosomes and show that most are inter
rupted with a regularly spaced AGG trinucleotide giving an ordered str
ucture to the array. Five types of arrays were identified consisting o
f varying numbers of a core unit with consensus [AGG(CGG)(9)]. Additio
nal variation in the length of the (CGG)(n) portion within each unit g
enerates the continuum of lengths seen on normal chromosomes. Ten per
cent contain long, uninterrupted tracts of (CGG)(n), and their lengths
suggest they have arisen by the loss of AGG triplets from longer inte
rrupted arrays. Haplotype analysis of arrays carrying long, uninterrup
ted (CGG)(n) tracts suggests that they occur more frequently on geneti
c backgrounds which are more highly represented on fragile X chromosom
es. These arrays may well be precursors from which the larger fragile
X associated arrays have arisen by further expansion.