1: \begin{abstract}

2: Using examples of the square- and triangular-lattice Heisenberg models

3: we demonstrate that the density matrix renormalization group method (DMRG) 

4: can be effectively used to study magnetic ordering in two-dimensional

5: lattice spin models. We show that local

6: quantities in DMRG calculations, such as the on-site magnetization $M$, 

7: should be extrapolated with the truncation error, not with its square root,

8: as previously assumed. 

9: We also introduce convenient sequences of clusters,

10: using cylindrical boundary conditions and pinning magnetic fields,

11: which provide for rapidly converging finite-size scaling.

12: This scaling behavior on our clusters is clarified using finite-size analysis 

13: of the effective $\sigma$-model and finite-size

14:  spin-wave theory.   

15: The resulting greatly improved extrapolations

16: allow us to determine the thermodynamic

17: limit of $M$ for the square lattice with an error comparable to

18: quantum Monte Carlo. For the triangular lattice, we verify the existence

19: of three-sublattice magnetic order, and estimate the order parameter to 

20: be $M = 0.205(15)$.

21: 

22: \end{abstract}

23: