Although one can prove it directly as in mercio's answer, it is insightful to deduce it as follows. Lagrange proved (1769) that in a group of odd order $\rm\,2n+1,\:$ every element is a square, since
$$\begin{eqnarray}\rm\, 1 &=&\rm a^{2n+1}\quad\ \text{by Lagrange's Theorem}\\ \rm \Rightarrow\ a &=&\rm a^{2n+2} = (a^{n+1})^2\end{eqnarray}$$
In a field $\rm\,\Bbb F\,$ of order $\rm\, q = 4n+3,\,$ the subgroup $\rm\,S\,$ of squares has odd order $\rm\,(q\!-\!1)/2 = 2n + 1,\,$ hence every square is a square (of a square), $ $ i.e. a fourth power; $ $ i.e. $\rm\ t \in S\:\Rightarrow\: t = s^2,\,\ s\in S,\ $ and, by definition, $\rm\:s \in S\:\Rightarrow\:s = a^2,\ a\in \Bbb F,\:$ hence $\rm\, t = s^2 = (a^2)^2 = a^4.$
Remark $ $ Lagrange's result is a special case of a $\rm\, k$'th power criterion that frequently proves handy. This criterion has a simple conceptual proof (alas, often overlooked). Suppose that we know that $\rm\, g^n = 1,\, $ so exponents on $\rm\,g\,$ can be interpreted $\rm\,mod\ n\!:\, i \equiv m\:$ $\Rightarrow$ $\rm\,g^i = g^{m}.\ $ We, $ $ consequently, $ $
see clearly that $\rm\,g^i\,$ is a $\rm\,k$'th power if $\rm\ mod\ n\!:\, k\mid i,\:$ i.e. $\rm\ i\equiv jk,\,$ so $\rm\,g^i = g^{jk} = (g^j)^k.\,$ By $\rm\color{#C00}{Bezout}$
$$\rm k\,|\, i\:\ (mod\ n)\!\iff\! \exists\,j\!:\ jk\equiv i\:\ (mod\ n)\!\iff\! \exists\, j,m\!:\ jk \!+\! mn = i\color{#C00}{\!\iff\!} (k,n)\,|\, i$$
Hence we have conceptually derived a proof of the following
Theorem $\rm\ \ \ g^n = 1,\,\ (k,n)\mid i\:\Rightarrow\: g^i\,$ is a $\rm\,k$'th power $\ \ $ [Easy $\rm\,k$'th Power Criterion]
Proof $\rm\ \ By\ Bezout,\,\ (k,n)\mid i\:\Rightarrow\:k\mid i\,\ (mod\ n) \Rightarrow\:i\equiv jk\,\ (mod\ n)\Rightarrow\: g^i = g^{jk} = (g^j)^k$
Note $\,\ $ That $\rm\ \ k\,|\, i\:\ (mod\ n)\!\iff\! (k,n)\,|\, i\ \ $ frequently proves conceptually handy, $ $ e.g. see here. $\ $ The reason behind this will become clearer when one studies cyclic groups and (principal) ideals.