Almost all normal-incommensurate phase transitions observed experimentally
are continuous. We show that there is not any theoretical reason for this g
eneral behavior in perfect crystals. A normal-incommensurate phase transiti
on that is not too far from the mean-field tricritical point should be disc
ontinuous and it is highly improbable that normal-incommensurate phase tran
sitions reported up to now lie very far from this point. To understand this
behavior we study the influence of defects on a hypothetical first-order n
ormal-incommensurate phase transition in a pure material. We have found tha
t this influence is strikingly different from that of other kinds of first-
order phase transitions. The change of the discontinuity of the order param
eter at the transition is negative and formally diverges within our approxi
mate theory. At the same time the diminishing of the phase-transition tempe
rature remains finite. We interpret these results as an indication that at
least some of the observed seemingly second-order normal-incommensurate tra
nsitions would be first-order transitions in defectless crystals.