1: \begin{abstract} 

2: A natural geometry, arising from the embedding into a Hilbert space 

3: of the parametrised probability measure for a given lattice model, 

4: is used to study the symmetry properties of real-space 

5: renormalisation group (RG) flow. In the projective state space this 

6: flow is shown to have two contributions: a gradient term, which generates

7: a projective automorphism of the state space for each given length scale;

8: and an explicit correction. We then argue 

9: that this structure implies the absence of any symmetry of a geodesic type

10: for the RG flow when restricted to the parameter space 

11: submanifold of the state space. This is demonstrated explicitly via a study 

12: of the one dimensional Ising model in an external field. In this 

13: example we construct exact expressions for the beta functions associated 

14: with the flow induced by infinitesimal rescaling. These constitute 

15: a generating vector field for RG diffeomorphisms on the 

16: parameter space manifold, and we analyse the symmetry properties of 

17: this transformation. The results indicate

18: an approximate conformal Killing symmetry near 

19: the critical point, but no generic symmetry of the 

20: RG flow globally on the parameter space.  

21: \end{abstract}