1 Introduction

A pioneering attempt to observe stellar spectra from outside the Earth's atmosphere was undertaken more than 35 years ago with the balloon-borne telescope named Stratoscope II, and water was clearly detected in Mira variables o Cet and R Leo (Woolf et al. 1964). Also, a possible presence of water in the normal M giants $\mu$ Gem (M3III) and $\rho$ Per (M4II) as well as in the early M supergiant $\alpha$ Ori (M2Iab) was suggested. The presence of water in such non-Miras, however, was so unexpected at that time (and even today) that it has not been understood correctly for a long time. Instead, the absorption bands at 1.4 and 1.9$\mu$m attributed to H₂O by the Stratoscope II observers were re-interpreted as due to the CN red system which also has the bandheads at 1.4 and 1.9$\mu$m (Wing & Spinrad 1970). In fact, it was known at that time that water can be observed only in the coolest M giant stars later than about M6 (e.g. Johnson & Méndez 1970), while CN can be well observed in the warmer red giant stars. Further support for the proposition by Wing & Spinrad (1970) was provided by the model photospheres of red giant stars developed at that time, which showed that water can be abundant in M giants with effective temperatures ( $T_{\rm eff}$) lower than about 3200K and that $T_{\rm eff}$'s of M giants should be revised upward against the ones known at that time (Tsuji 1978). This upward revision of the effective temperature scale was well consistent with the empirical scale based on the angular diameter measurements which showed $T_{\rm eff} \approx$ 3250K for M6III (Ridgway et al. 1980). Thus, theory and observations that water can be observed only in M giants later than about M6 appeared to be consistent. This result further made it difficult to accept the Stratoscope II result that water could be detected in the M giant stars as early as M3-4.

Nevertheless, we recently found the possible presence of water in the early M giant $\beta$ Peg (M2.5II-III) (Tsuji et al. 1997) on the spectra we observed with the Short Wavelength Spectrometer, SWS (de Graauw et al. 1996), on board the ISO (Kessler et al. 1996). This result was based on the analysis of the 2.7$\mu$m region where the H₂O $\nu_1$ and $\nu_3$ bands can be found, but the overlapping OH and CO bands made it difficult to clearly demonstrate the presence of the H₂O bands, especially by the low resolution spectrum we had at that time. Also, possible presence of water in early M type stars was suggested by the low resolution data obtained with IRTS (Infrared Telescope in Space) of ISAS (Matsuura et al. 1999). By these results, however, it might still be difficult to convince the presence of water in non-Mira stars earlier than about M6 against the general belief that water should not exist in such stars. Now, it is possible to utilize a larger sample of high resolution ISO spectra recently released by ESA and, with the higher resolution, we detected the H₂O $\nu_2$ bands in the 6.3$\mu$m region, where is little disturbed by other molecular bands. This observation finally provides convincing evidence for the presence of water in normal red giants including the late K and early M giant stars.

 

 

Table 1: Program stars observed with the ISO SWS.

Nos.	object	BS	Sp. type	$T_{\rm eff}$(K)	ISO Obsno.
1	$\alpha$ CMa	2491	A1Vm	$\approx\!\!10\,000$	689 01202
2	$\alpha$ Boo	5340	K1IIIb	4362a	452 00101
3	$\gamma$ Dra	6705	K5III	$4095\pm163^{b}$	040 02405
4	$\alpha$ Tau	1457	K5III	3898a	636 02102
5	$\beta$ And	337	M0IIIa	$4002\pm178^{b}$	795 01002
6	$\alpha$ Cet	911	M1.5IIIa	$3869\pm161^{b}$	806 00924
7	$\beta$ Peg	8775	M2.5II-III	$3890\pm174^{b}$	551 00705
8	$\gamma$ Cru	4763	M3.5III	3626a	609 00804

$^{{\rm\it {a}}}$

Cohen et al. (1996). $^{{\rm\it {b}}}$ Dyck et al. (1998).