3 Results and discussion

As said before we have observed 137 stars designated as CS or candidates. A first examination permits to separate 22 stars which have a spectrum of a hot star of type B, A, Am or F in the near infrared. These stars are listed in Table 1. The other 115 stars have a cool spectrum of types G, K or M. These are listed in Table 2, except 20 stars for which the classification from Aurelie and from Carelec coincide totally (see part 3a). In the two tables the data V and B-V were taken from the Hipparcos and Tycho catalogues (1997). If they are not contained in these catalogues, the data come from the Centre de Données Stellaires (CDS) or from Hynek (1938). An asterisk implies a note at the end of the table. In the column "references", for reasons of space economy, figures only the name of the first author of the paper.

We next will validate our classifications obtained from Aurelie with proceeding from Carelec. Then we shall confront our classifications with those of the "standard" CS, that is, stars which possess a classification which we consider trustworthy (see Paper II, 4.1). And finally we shall confront, for a small sample, our results with those published in the Michigan catalogues (Houk 1982; Houk & Smith-Moore 1988).

a) Comparison Carelec-Aurelie
A comparison of the 51 stars observed with both spectrographs permits to isolate 22 stars for which the results are identical. These are: HD 4775 (G7III) - 17878 (G7III) - 21771 (K3II) - 23089 (G2Ib) - 25555 (K1II) - 26630 (G0Ib) - 26673 (G9II) - 29094 (G7Ib) - 39118 (K0II) - 66094 (G9III-IV) - 74395 (G1Ib) - 88021 (G8III) - 166479 (G8III) - 169985 (G9III) - 184759 (G8III) - 187259 (G8III) - 193410 (K0III) - 193495 (G9II) - 197177 (G8II) - 200428 (K0III) - 205114 (G2Ib) - 213310 (K6-M0I). These objects do not figure in Table 2, except HD 4775 and HD 25555 for which recent classifications exist, obtained on the basis of the method of substraction of spectra (Griffin R&R 1986). For the other 29 stars one perceives small differences in spectral type and luminosity classes (Fig. 4), but without a systematic trend. The average scatter around the linear relations corresponds to about 0.7 tenth of a spectral type and 0.3 luminosity class. Both errors are within the normal precision of spectral types (Jaschek & Jaschek 1987).

b) Comparison with the so called "standards" CS
We have made the inventory of CS stars for which a satisfactory classification exists (see Paper II). The complete list is as follows: HD 4775, 17878, 25555, 26630, 29094, 57146, 74395, 88021, 169689, 173764, 184759, 187076, 187259, 187299, 190161, 192713, 200428, 200905, 201270, 205114.

Of this list 13 have classifications obtained by R&R Griffin by the so called method of substraction of spectra and six others are MK standards.

  

Table 1: Stars of the sample which exhibit a hot spectrum

In Fig. 5 we compare our Aurelie classifications with those of the "standards" for spectral types and for luminosity classes respectively.

  

Table 2: Classifications of the cool components, SP(IR), of stars with composites spectra

  

Table 2: continued

  

Table 2: continued

From the dispersions around the average relation we find again values which are normal for the errors of classification (one tenth of spectral type and 0.25 luminosity class). Also there does not exist a systematic effect.

c) Comparison with the classification of the Michigan catalogues
Our complete sample (Papers II and III) contains 25 stars of the southern hemisphere which figure in the catalogues of Michigan: HD 29961, 35162, 47579, 51250, 57146, 59076, 60414, 70442, 74946, 75098, 79267, 84367, 95235, 102171, 120901, 137975, 167570, 168701, 169652, 173805, 174191, 193495, 209278, 218640, 223932.

  

Figure 5: Comparison of infrared MK classifications (spectral types and luminosity classes) with those of "standard composite spectra" (see text). Sp T: G...1, K...2, M...3; Lumi.: Ia...0, Ib...1, II...2, III...3; numbers on the plots are weights

  

Figure 6: Comparison of infrared MK classifications (spectral types and luminosity classes) with those of N. Houk (Michigan Catalogues). Sp T: G...1, K...2, M...3; Lumi.: Ia...0, Ib...1, II...2, III...3; numbers on the plots are weights

In Fig. 6 we compare their and our classifications, for spectral types and luminosity classes respectively. We notice that for HD 209278 Houk does not provide the luminosity classification.

The spectra classed by Houk were obtained with the objective prism, on IIaO plates, dispersion 108 Å/mm at H$_\gamma$ and a resolution near to 2 Å. If one takes into account the difficulties in classifying CS in this wavelength region (see Paper I), the agreement with our classifications is very good for the spectral types. On the average, the Michigan classifications are slightly later than ours for type G and slightly earlier for type K. For luminosity classes the agreement is excellent for class III, but in disagreement for classes I and II. From five stars classed II by us, four are classed III by Houk and two stars classed I by us are classed II and III by Houk.