We have studied the multiple recording method to estimate if it possible to
obtain by this way reliable EXAFS data standard deviations. The examinatio
n of more than 10 different samples, each of them recorded at least 8 times
, gave us the possibility to show that the partial standard deviations foll
ow the expected probability laws. Even when the number of samples is small,
it is possible to give a reasonable upper limit of epsilon which could be
used as the statistical contribution of the experimental error bars. In som
e cases we have found that a uniform weighting by 1/epsilon(2) may fail to
give a good fit, especially in the high part of k space. We present a simpl
e method of weighting which lead to more satisfying fits without loosing th
e correct normalisation of the fined Delta chi(2) necessary to obtain corre
ct fitted parameters error bars.