Neutrino mixing matrix appears to be close to bimaximal mixing, but for the solar mixing angle which is definitively smaller than forty five degrees. Whereas it seems quite easy to understand bimaximal mixing with the use of new global symmetries, as in models using $L_e - L_��- L_��$, understanding the about to eleven degrees of deviation in the observed solar angle seems less simple. We suggest that such a deviation could be due to a light sterile neutrino that mixes with the active sector. The mass scale needed to produce the effect has to be smaller than atmospheric scale, and it would introduce a new mass squared difference which should be smaller than the solar scale. We present a toy model that exemplifies these features.

19 pages, two figures. Discussion extended. References added