The bright star Epsilon Aurigae (HD 31964) is a single-lined spectroscopic binary that is famous for its long orbital period (27.1 yr), which is punctuated by an almost two-year long eclipse caused by an essentially invisible object.(from Hoard D.W., Howell S.B. & Stencel R.E. (2010). Taming the invisible monster: Systme parameter constraints for ϵ Aurigae from the far-ultraviolet to the mid-infrared, The Astrophysical Journal, 714 (1) 549-560. DOI: 10.1088/0004-637X/714/1/549
By examining the optical spectra of Epsilon Aurigae near the end of its 1954-1956 eclipse, Hack(1) was able to deduce the electron density and develop the hypothesis of a Be-star-like hot object at the center of a large disk of occulting material(2).
(1) Margherita Hack compared her observations with Adams and Sanford (1929, published the next year):
The spectrum shows a doubling of the lines similar to that in 1929, with the violet component appearing to originate in a very rarefied shell, which surrounds an invisible star in accordance with Struve's hypothesis (1956) and is responsible of the observed eclipse. The study of the shell spectrum by means of the curve of growth shows dilution of radiation effecfts.Hack M. (1959). The Spectrum of Epsilon Aurigae., The Astrophysical Journal, 129 291. DOI: 10.1086/146620
(2) Some of the reported observations were subsequently revised by other researchers, but the road was basically that
several evidences consistently suggest that the invisible companion is a hot star responsible for the excitation of a gaseous shell surrounding it. This shell is the body responsible for the eclipse of the principal star. The invisible companion probably is a P Cygni star of absolute visual magnitude -4 or -5, and effective temperature of 20000 or 30000 K.Hack M. (1962). A new explanation of the binary system $\varepsilon$ Aurigae, Memorie della Società Astronomia Italiana, Vol. 32, p.351
About the shell, with Ferluga, she observed that
this shell is responsible for selectively absorbing the primary's light and producing the "shell" spectrum.Ferluga S., Hack M. (1985). High-dispersion spectroscopy of the eclipse of Epsilon Aurigae at visible and ultraviolet wavelengths, Astronomy and Astrophysics, vol. 144, no. 2, March 1985, p. 395-402.
We may wonder how it is possible for the gaseous, partially ionized shell to coexist with the cool object. We think that the light of the hot secondary escaping from the poles is able to excite and ionize the thin gaseous shell which extends farther out of the orbital plane, while a cool disk or ring surrounding the secondary and eclipsing the primary lies in vicinity of the orbital plane, and makes the secondary appear fainter than it really is.