Geologic structures are mostly known from scattered data, and structur
es such as folds or faults are drawn in between by using interpolation
, which is often based on geologically poor assumptions, such as smoot
hness. The need for more accuracy leads to restoration techniques in w
hich more realistic assumptions are introduced. In this context we hav
e tested a multisurface unfolding procedure. We use a least-squares fo
rmulation involving the following criteria: initial horizontality, bed
-length conservation (during slip on bedding) and local volume conserv
ation. Weighted optimization of these criteria gives a compromise betw
een them if they are conflicting. We have successfully tested the meth
od on various theoretical examples and on an analog model: the 'paperb
ack experiment'. (C) 1997 Elsevier Science Ltd.