1: \begin{abstract} It is well known that every subqadratic Dehn function is linear. A question by Bridson asked to describe the isoperimetric spectrum of groups, that is the set of all numbers $\alpha$ such that $n^\alpha$ is equivalent to the Dehn function of a finitely presented group.

2: The goal of this paper is to give a description of the isoperimetric spectrum. Earlier a similar description was given by Sapir, Birget and Rips for the intersection of the isoperimetric spectrum with $[4,\infty]$. Lowering the bound from 4 to 2 required significant new ideas and tools.

3: %On the one hand, it is well known that %the only subquadratic Dehn function of %finitely presented groups is the linear %one. On the other hand there is a huge %class of Dehn functions $d(n)$ with %growth at least $n^4$ (essentially all %possible such Dehn functions) constructed %in

4: %\cite{SBR} and based on the time %functions of Turing machines and

5: %S-machines. The class of Dehn functions %$n^{\alpha}$ with $\alpha\in (2; 4)$ %remained more mysterious even though it %has attracted quite a bit of attention %(see, for example, \cite{B}, \cite{BB}, %\cite{BBFS}).

6: %We fill the gap by obtaining Dehn %functions of the form $n^{\alpha}$ (and %much more) for all real $\alpha\ge 2$ %computable in reasonable time, which is %defined as in \cite{SBR}. This completes %the description of the spectrum of Dehn %functions of the form $n^{\alpha}$ modulo %the hypothesis $\bf P=NP$. For example, %$\alpha$ can be $\pi$ or $e$, or any %algebraic number greater than $2$. As in %\cite{SBR}, we use S-machines but newly %introduced tools and proofs are %essentially used for the best possible %lower bound $d(n)\ge n^2$.

7: \end{abstract}

8: