Recently, a method for the identification of multiphonon states in vib
rational nuclei was successfully developed. A survey was performed and
it was found, surprisingly, that the energies of the first 2(+) state
s are systematically low by about 15% with respect to the values expec
ted from the global nuclear properties. This appears to be in contradi
ction with the general belief that these states have a high purity. It
can be questioned whether such results are not simply an artefact due
to the weights used in the fits. One way to answer this question is t
o compare the experimental results with the predictions of the Brink m
odel. To this end, the equations for four and five phonons were explic
itly calculated. It will be shown that the predictions are quite good,
provided the one phonon energy be renormalized, i.e. increased by abo
ut 15%. Since the modified Brink method involves only the use of a ''v
irtual'' 2(1)(+) energy and no level fit, a problem of weights cannot
be invoked. Possible explanations of the anomaly are examined. (C) 199
7 Elsevier Science B.V.