1: \begin{abstract}

2:   The question of the transition to global isotropy from our

3:   anisotropic local universe is studied using the Union~2 catalogue of

4:   Type~Ia supernovae (SNe\,Ia).  We construct a `residual' statistic

5:   sensitive to systematic shifts in their brightness in different

6:   directions and use this to search in different redshift slices for a

7:   preferred direction on the sky in which the SNe\,Ia are brighter or

8:   fainter relative to the standard $\Lambda$CDM cosmology. At low

9:   redshift ($z<0.05$) we find that an isotropic model such as

10:   $\Lambda$CDM is barely consistent with the SNe\,Ia data at 2--3

11:   $\sigma$. A maximum likelihood analysis of peculiar velocities

12:   confirms this finding --- there is a bulk flow of 260~km\,s$^{-1}$

13:   extending out to $z\sim0.06$, which disagrees with $\Lambda$CDM at

14:   1--2 $\sigma$. Since the Shapley concentration is believed to be

15:   largely responsible for this bulk flow, we make a detailed study of

16:   the infall region: the SNe\,Ia falling away from the Local Group

17:   towards Shapley are indeed significantly dimmer than those falling

18:   towards us on to Shapley. Convergence to the CMB rest frame must

19:   occur well beyond Shapley ($z>0.06$) so this low redshift bulk flow will

20:   systematically bias any reconstruction of the expansion history of

21:   the universe. At higher redshifts $z>0.15$ the agreement between the

22:   SNe\,Ia data and the $\Lambda$CDM model does improve, however, the

23:   sparseness and low quality of the data means that the latter cannot

24:   be singled out as the preferred cosmological model.

25: \end{abstract}

26: