Width of the Confining String

We critically reconsider our recent observation of confining string shrinkage in pure glue SU(2) lattice gauge theory near the continuum limit. Using the advanced numerical techniques we argue that imperfect overlap with the string ground state and corresponding ambiguities in T->\infty extrapolation are likely to be the cause of the observed scaling violations. In particular, even in the limit of large Euclidean times the string shrinkage is apparent in torelon correlation function, however, in this case the best relevant overlap we could attain is of order 50%. To the contrary, when the ground state is selected properly the string width scales in physical units being ~0.3 fm for 1 fm long confining string.