1 Introduction

MG 2016+112 is one of the first discovered gravitational-lens system. Lawrence et al. (1984) observed three distinct radio sources, A, B, and C. Further higher resolution radio map resolved C into multiple components (Garrett et al. 1994, 1996). At optical wavelength, the image A and B are point-sources but the image C is a fainter resolved object (Lawrence et al. 1984; Schneider et al. 1985, 1986). Spectroscopic observations revealed that A and B show very similar spectra dominated by strong ultra-violet (UV) emission-lines redshifted to z=3.27 (Lawrence et al. 1984; Schneider et al. 1987). Although the optical spectrum of C has not been fully published so far, it has always been considered at the same redshift as A and B after its detection in the redshifted Ly$\alpha$ narrow-band data (Schneider et al. 1986).

The MG 2016+112 lens system is enigmatic for 3 reasons. First, the lensed object itself is very unique among known high-redshift galaxies (Lawrence et al. 1984). It has been conventionally called a "quasar" since it is a fairly luminous point-like object. However, A and B show only narrow ultra-violet (UV) emission lines and thus MG 2016+112 is not an ordinary broad-line quasar. Although strong narrow emission lines are typically seen for high-redshift powerful radio galaxies (HzPRGs), MG 2016+112 appears somewhat different from the known HzPRGs. Indeed, while HzPRGs are typically extended in optical images and have lobe-dominated radio structures with a scale of at least a few tens of kpc, A and B show only point-like features even in high-resolution optical and radio images. It is important to investigate the true nature of the lensed object.

Second, image C is a very complex object: at optical and near-infrared (NIR) wavelength, C is resolved and has an arc-like morphology. Radio-to-optical flux ratio of C is several times larger than those of A and B and cannot be explained by variability and time delay. At radio wavelength, C is resolved into two components, C₁ and C₂. While A, B and C₂ (sometimes referred as C$^\prime$ in literatures) are point-like objects even at 15-mas resolution, C₁ has been further resolved into three chain-like components (Garrett et al. 1996). Radio spectral shape of C₂ is similar to those of A and B ( $\alpha = 0.8$) but C₁ has significantly flatter one ( $\alpha = 0.2$). Whether C₁ is a radio galaxy at different redshift or a lensed image of the outer structure of the radio source at z=3.27 is still in question. Finally the nature of the lens producing this multiple image system has been enigmatic. Deep optical and NIR images have detected a red galaxy D amid images A, B and C (Schneider et al. 1986; Langston et al. 1991; Lawrence et al. 1993). Galaxy D seems to correspond to an evolved giant elliptical galaxy at $z\sim 1$. However, the image separation requires a mass-to-light ratio for D much larger than typical one. While lens models have assumed the existence of a high-redshift massive cluster as an additional source of lensing mass (Narasimha et al. 1984, 1987, 1989; Nair & Garrett 1997; Langston et al. 1991), at first no signature of such a cluster has been seen in the optical and NIR observations (Schneider et al. 1987; Langston et al. 1991).

Hattori et al. (1997) detected an extended X-ray emission in the direction of MG 2016+112, possibly emitted by the hot gas in the lensing cluster of galaxy. A strong emission-line-like feature consistent with iron lines redshifted to z $\sim 1$ was detected. This discovery could in principle solve the "dark-lens" problem for MG 2016+112, but opened another question of a "dark cluster", namely, the lack of optical counterpart of the X-ray hot gas. Very recently, Soucail et al. (2000) and Kneib et al. (1997) spectroscopically detected several galaxies at $z\sim 1$ and Benítez et al. (1999) observed a possible color-magnitude sequence of faint red galaxies in this field arguing for the existence of a distant massive structure at $z\sim 1$.

We have obtained new spectra of components B, C, D of the MG 2016+112 system. In this paper, we analyze the emission lines of image B and C in order not only to investigate the intrinsic nature of MG 2016+112 but also to understand the component C in view of the lens-model prediction. Observations and data reduction are described in Sect. 2. In Sect. 3, we examine the observed emission line properties. The line flux ratios are compared with the prediction of the photo-ionization models as well as those of other high-redshift objects in various categories. In Sect. 4, the nature of image C is discussed in the context of lens models. We then argue that MG 2016+112 may be a radio-quiet obscured quasar based on the results obtained in Sect. 3 as well as those discussed in literatures. Throughout this paper we use H₀=50 kms^-1/Mpc, $\Omega_0=1$ and $\lambda=0$.