# If SLLN holds, how can I say sum of variances is finite

So SLLN holds and mean of the independent random variables $xi$ is 0, I have to prove that the series of $\frac{\xi_n}{n^{1+\alpha}}$ converges for any $\alpha > 0$.

I'm using 3-series theorem but not sure how to use SLLN to prove that the series $\operatorname{Var} \left( \frac{\xi_n}{n^{1+\alpha}} \right)$ converges. I'd greatly appreciate your help.

-

Introduce $$\eta_n=\frac1n\sum_{k=1}^n\xi_k\qquad\text{and}\qquad a_n=n\left(\frac1{n^{1+\alpha}}-\frac1{(n+1)^{1+\alpha}}\right).$$ Then, $$\sum\limits_{n\geqslant1}\frac{\xi_n}{n^{1+\alpha}}=\sum\limits_{n\geqslant1}a_n\eta_n.$$ The proof is complete, using the following facts:
1. The sequence $(\eta_n)_{n\geqslant1}$ is almost surely bounded since $\eta_n\to0$ almost surely.
2. The series $\sum\limits_na_n$ converges since $a_n\geqslant0$ for every $n$ and $$\sum\limits_{n\geqslant1}a_n=\sum\limits_{n\geqslant1}\frac1{n^{1+\alpha}}\leqslant1+\int_1^{+\infty}\frac{\mathrm dx}{x^{1+\alpha}}=1+\frac1{\alpha}.$$
@Darwin, No. Hint: pretend for a moment that $\xi_n=1$ for every $n$, hence $\eta_n=1$ for every $n$. Then look at the formula which expresses the $\xi$ series as an $\eta$ series... – Did Dec 20 '11 at 8:10
There is a factor $n$ in the numerator of $a_n$ AND the sum of the series $\sum\limits_{n\geqslant1}a_n$ is what I wrote. (Did you try my hint?) – Did Dec 20 '11 at 16:21