poses Bd ele ote Bee Da wo ee at a outermost layers of the model. In this case however. one may be certain that the equation of transfer that has been adopted to represent line formation is star different from the one seen in a is not appropriate to calculate degree of ionization and level of excitation by the equilibrium relations known as Saha’s and Boltzmann’s laws. In solv- the brightness at 4500 A, is due to the no longer tenable: furthermore, that it ing the problems of the monochromatic transfer of radiation through the outer parts of the stellar atmosphere one should take into account what are called “non-lt.e. effects’—that is, departures from the equations represent- ing local thermodynamic equilibrium. In order to obtain a first prediction brightness (Fig. the spectrum has been predicted by use of the existing thermal-equilibrium theory of line formation. When observed line profiles are compared with predicted line profiles, attention must be paid to isolation of discrepancies due to (i) de- partures from the adopted simple theory of line formation: (ii) departures of the state of motion, of the gas in the plane parallel layers, from the thermal motions appropriate to the local temperature, which are the only motions bly deeper layers than does the spectrum Expected Ultraviolet Absorption Lines tion of the depth of an absorption- The ultraviolet spectrum of O and B stars divides into two regions. In the section 911.6 to 1900 A there at 4500 A. Important in determina- line profile at any spectral region is the factor of whether the selected frequency range lies on the steep side the stronger mosphere, and where it occurs, is the “depth of formation” of the spectrum at frequency ». Whether this frequency lies in the continuous spectrum, Or in shown that Fp, the emergent flux in frequency y, is approximately equal to By, the value of the Planck function at this frequency, at the layer in the model in which the monochro- matic optical depth t) is about equal to 0.4. In a very opaque spectral region the position at which ftequals 0.4 will occur in the outermost layers of the star; in a transparent spectral region fy will equal 0.4 at much greater depths. In the first case, the local temperature will be low; in the second it may be higher by 10,000° or 15.000°K. Clearly, in an opaque spectral region such as the centre of a strong line, one is observing a part of the 8 DECEMBER 1967 lines are not deep; at only important spectra not having reso- nance lines in this region are those of He I and H, O II and Il, and Mg II. Morton dicted the profiles of stronger lines formed in cially for resonance lines, where line- formation sketched above. The resuits that the presence of strong lines in the in the neighborhood of 1200 A appear in Fig. 3; the great width and depth ultraviolet spectral region causes greater reduction of the expected emergent flux, from the flux for the case of no a We ee es ee ee Fy ~ of the resonance lines are evident. The central parts of these lines are formed ee e ee e ee ee es UNBLANKETED rh. ' ee ee ee " ———= MODEL ---- . TYPE B15 rs it ee PREDICTED SPECTRUM LINE-BLANKETED MODEL 4o0b IH, Ne I and (4) has presome of the an unblanket- ed model atmosphere of type about BI.5, using the simple theory of line formation by scattering processes must be important), but it does demonstrate ‘ I J -_ hI 3.0} ~ ’ { a line, is immaterial; what is significant is that numerical work has second, and third ions of the elements He to Ca; many of these lines are predicted to be exceedingly strong. The cially at very short wavelengths. The foregoing qualitative explanation does not tell the full story (espe- An important concept for visualiza- tion of what occurs in the stellar at- lines from low-lying levels of the first, tail. In the latter case (at wavelengths (iii) the fact that the stellar atmo- metric form. occur Many resonance lines and strong of the Planck function for the relevant temperatures or on the long-wavelength taken into account in the theory: and sphere may indeed be so extensive that it cannot be adequately represented by a plane parallel layer, but must be represented as a sphere or other geo- shown in Fig. 2. trum at 1300 A comes from considera- wavelengths shorter than 3000 A, the strong lines become very deep, espe- and factors causes fact that in these calculations the spec- erations have ignored play of these various the two different spectral distributions 1), in comparison to greater than 3600 A in B stars) even been tween 4000 and 5000 A. The inter- lines. A large part of the ultraviolet of the ultraviolet spectrum from OB stars, these important physical consid- lines, than would ever occur in the normally observed spectral region be- transparent spectral region, such as in the continuous spectrum or in weak t — 4 — i [ t 2.0+! i i ~ + - ' olla | BALMER JUMP ee - 1000 | of poy tt 2000 lg tl ng 3000 4000 FG 5000 yl tl § 000 Wavelength (A) Fig. 2. The difference in spectral distribution between a line-blanketed model calculated by C. Guillaume and an unblanketed model calculated by one of us (A.B.U.). The ynjt of flux is 10“ erg per square centimenter per second per unit frequency interyayy:, {275