1: \begin{abstract}

2: 

3: In a recent paper (Phys.\ Rev.\ B {\bf 59}, 9699 (1999)), Chandross and 

4: Hicks claim to present a new density matrix renormalisation group (DMRG) 

5: method for dealing with excited states of quantum lattice models. The 

6: proposed improvement to the DMRG---the inclusion of excited state wave

7: functions 

8: {\em in addition} to the ground state in the density matrix when calculating

9: excitations---is in fact standard pratice, is clearly stated in White's 

10: original papers, and has been used repeatedly by many groups to study 

11: excited states. The authors apply the method to the extended, dimerised 

12: Hubbard model for conjugated polymers. The criteria for determining whether 

13: states are bound or not are assessed. The authors claim that their results 

14: show that the optically important ``$1B_u$'' state is bound (excitonic), in 

15: contrast to a previous study. However, the discussion is qualitative, and 

16: the authors arrive at conclusions on the basis of results for one lattice 

17: size only. We show that when Chandross and Hicks' criterion is developed 

18: into a quantitative definition of particle-hole separation, with the 

19: finite-size dependence analysed, the implication is that the 

20: $1B_u$ state is unbound, in keeping with the conclusions of a previous 

21: study.

22: 

23: \end{abstract}

24: