1: \begin{abstract}

2: It is believed that, in the limit as the conductor tends to infinity, correlations between the zeros of elliptic curve $L$-functions averaged within families follow the distribution laws of the eigenvalues of random matrices drawn from

3: the orthogonal group.  For test functions with restricted support, this is known to be the true for

4: the one- and two-level densities of

5: zeros within the families studied to

6: date. However, for finite conductor Miller's

7: experimental data reveal an interesting discrepancy from

8: these limiting results. Here we use the $L$-functions ratios

9: conjectures  to calculate the 1-level density for the family of

10: even quadratic twists of an elliptic curve $L$-function for large but finite conductor.  This

11: gives a formula for the leading and lower order terms up to an error term that is conjectured to be significantly smaller. The lower order terms explain many of

12: the features of the zero statistics for relatively small conductor and

13: model the very slow convergence to the infinite conductor limit.

14:  However, our main observation is that they do not

15:  capture the behaviour of zeros in the important region

16:  very close to the critical point and so do not explain Miller's discrepancy.   This therefore implies

17:  that a more accurate model for statistics near to this point

18:  needs to be developed.

19: \end{abstract}

20: