1: \begin{abstract}

2: M star spectra,  at wavelengths beyond 1.35 $\mu$m, 

3: are dominated by water vapour yet

4: terrestrial water vapour makes it notoriously difficult to make accurate

5: measurement from ground-based observations.

6: We have used the short wavelength spectrometer on the Infrared Space

7: Observatory at four wavelength

8: settings to cover the 2.5--3.0 $\mu$m

9: region for a range of M stars. The observations show a good match

10: with previous ground-based observations and with 

11: synthetic spectra based on the Partridge \& Schwenke (1997) line list 

12: though not with the SCAN (Jorgensen et al. 2001) line list.

13: We used a least-squared minimisation technique to systematically

14: find best fit parameters for the sample 

15: of stars. The  temperatures that we find indicate a relatively hot 

16: temperature scale for M dwarfs.

17: We consider that this could be a 

18: consequence of problems with the Partridge \& Schwenke linelist which leads

19: to synthetic spectra

20: predicting water bands which are too strong for a given temperature.

21: Such problems need to be solved in the next generation 

22: of water vapour line lists which will extend the calculation

23: of water vapour to higher energy levels with the good 

24: convergence necessary for reliable modelling of hot water vapour.

25: Then water bands can assume their natural role as the primary tool

26: for the spectroscopic analysis of M stars.

27: 

28: \end{abstract}

29: