1: \begin{abstract}

2: A recently proposed molecular model is discussed as a non-trivial extension

3: of the Ising model. For $d=2$ the two models are shown to be equivalent,

4: while for $d>2$ the molecular model describes a peculiar second order

5: transition from an isotropic high temperature phase to a low-dimensional

6: anisotropic low temperature state. The general mean field analysis is

7: compared with the results achieved by a variational Migdal-Kadanoff

8: real space renormalization group method and by standard Monte Carlo

9: sampling for $d=3$. By finite size scaling the critical exponent has

10: been found to be

11: $\nu=0.44\pm 0.02$ thus establishing that the molecular model

12: does not belong to the

13: universality class of the Ising model for $d>2$.

14: \end{abstract}

15: