1: \begin{abstract}

2: A Cram\'{e}r moderate deviation theorem for Hotelling's

3: $T^{2}$-statistic is proved under a finite $(3+\delta)$th moment. The

4: result is applied to large scale tests on the equality of

5: mean\vspace*{1pt} vectors and is shown that the number of tests can be

6: as large as $e^{o(n^{1/3})}$ before the chi-squared distribution

7: calibration becomes inaccurate. As an application of the moderate

8: deviation results, a global test on the equality of $m$ mean vectors

9: based on the maximum of Hotelling's $T^{2}$-statistics is developed and

10: its asymptotic null distribution is shown to be an extreme value type I

11: distribution. A novel intermediate approximation to the null

12: distribution is proposed to improve the slow convergence rate of the

13: extreme distribution approximation. Numerical studies show that the new

14: test procedure works well even for a small sample size and performs

15: favorably in analyzing a breast cancer dataset.

16: \end{abstract}