On the interpretation of S Doradus.

Three slit spectra of S Dor during the spectroscopic minimum revealed in the years 194546 are compared with two taken at maximum phase in the years 1952 and 1957 respectively. At the minimum phase (which is well marked in 1946, February 2)14 [Fe ii] emission lines are recorded between 3900 and 5000 A, compared with 7 observed by Wesselink in the minimum of 1955 and 20 observed by Thackeray in the minimum of 1964. [S H] is also represented by the line at 4068 A. Mg H 4481 appears to be weakly present in absorption while He i (which Wesselink and Thackeray found relatively strong) is absent. Fe H is represented by 14 emission lines; very few of these lines show a P Cyg structure: on February 2, 1946, only Fe II 4583 shows a companion absorption line. Ti H is absent both in emission and in absorption. The spectral type suggested is B8eq. At the maximum phase, the P Cyg features are very well developed in the lines of H, Fe H, Ti H, Ca H. The Ti -Fe H anomaly is well marked in 1952 but not in 1957. The spectral types suggested are A5eq in 1952 and A3eq in 1957. All the lines observed in the range 4861-3691 A have been identified and a visual estimate of their intensity at the various phases is given. The radial velocity has been measured for all the unblended lines. The emission lines show a shift appreciably constant at the various phases while the absorption lines are more shortward displaced at maximum than at minimum. By combining the photometric and spectroscopic observations, the general behaviour of S Dor is examined and possible explanations are discussed. It seems that an ejectedshell mechanism is not inconsistent either with the observational evidences or theoretical requirements. From the evolutionary point of view, the possibility that S Dor is a contracting pre-main-sequence star is considered. Key words: S Doradus - observations - spectroscopic behaviour - instability - evolution