$\begin{figure} \par\includegraphics[height=5.7cm,angle=0,clip]{Da311_f1.ps}\includegraphics[height=5.7cm,angle=0,clip]{Da311_f2.ps}\end{figure}$

Figure 1: Finding charts for ISO J1324-2016. The images are 45'' on a side; North is up, East is left. Left: ESO NTT/SUSI I image (exp. time: 1800 s). Right: ESO VLT/ISAAC K acquisition image (exp. time: 40 s).

$\begin{figure} { \includegraphics[width=12cm,angle=0,clip]{Da311_f3.eps} } \end{figure}$

Figure 2: Right: portion of the ISAAC H-band spectral interval showing the H $\alpha$ line at a redshift of 1.50. The original spectrum has been lightly filtered by a Gaussian with a $\sigma$ of 1 pixel (3.6 Å). Intensity is in arbitrary units. Left: portion of the ISAAC J spectral interval containing the H $\beta$ and [OIII] lines. No line is detected in the K-band spectrum. In each panel, the lower curve is the noise spectrum extracted from the same data.

Table 1: Photometry of ISO J1324-2016. $B,~ R,~ I~ \&~ K$ flux densities have been corrected for galactic extinction (NED).
Observed Rest Flux Telescope/

Wavelength Wavelength Density Instrument

$B_{\rm J}$    0.44 $\mu$ m 0.176 $\mu$ m <0.98 $\mu$ Jy CFHT (1993)

$R_{\rm J}$    0.70 $\mu$ m 0.28 $\mu$ m <2.8 $\mu$ Jy CFHT (1993)

$I_{\rm J}$    0.90 $\mu$ m 0.36 $\mu$ m 3.0 $\pm ~ 0.3$ $\mu$ Jy NTT/SUSI (1997)

$K_{\rm J}$    2.2 $\mu$ m 0.88 $\mu$ m 67 $\pm ~6$ $\mu$ Jy UKIRT (1998) & VLT/ISAAC (2000)

LW2 6.75 [5-8.5] $\mu$ m 2.7 $\mu$ m 0.89 _-0.33^+0.47 mJy ISOCAM (1996)

LW3 15 [12-18] $\mu$ m 6.0 $\mu$ m 0.76 _-0.40^+0.87 mJy ISOCAM (1996)

13 cm 5.0 cm 0.8 $\pm$ 0.10 mJy ATCA (1995)

22 cm 8.9 cm 1.4 $\pm$ 0.12 mJy ATCA (1995)

35 cm 14.2 cm <2.1 mJy MOST (1993)

**Table 1:** Photometry of ISO J1324-2016. $B,~ R,~ I~ \&~ K$ flux densities have been corrected for galactic extinction (NED).
Observed	Rest	Flux	Telescope/
Wavelength	Wavelength	Density	Instrument
$B_{\rm J}$ 0.44 $\mu$ m	0.176 $\mu$ m	<0.98 $\mu$ Jy	CFHT (1993)
$R_{\rm J}$ 0.70 $\mu$ m	0.28 $\mu$ m	<2.8 $\mu$ Jy	CFHT (1993)
$I_{\rm J}$ 0.90 $\mu$ m	0.36 $\mu$ m	3.0 $\pm ~ 0.3$ $\mu$ Jy	NTT/SUSI (1997)
$K_{\rm J}$ 2.2 $\mu$ m	0.88 $\mu$ m	67 $\pm ~6$ $\mu$ Jy	UKIRT (1998) & VLT/ISAAC (2000)
LW2 6.75 [5-8.5] $\mu$ m	2.7 $\mu$ m	0.89 _-0.33^+0.47 mJy	ISOCAM (1996)
LW3 15 [12-18] $\mu$ m	6.0 $\mu$ m	0.76 _-0.40^+0.87 mJy	ISOCAM (1996)
13 cm	5.0 cm	0.8 $\pm$ 0.10 mJy	ATCA (1995)
22 cm	8.9 cm	1.4 $\pm$ 0.12 mJy	ATCA (1995)
35 cm	14.2 cm	<2.1 mJy	MOST (1993)

Radio data
The cluster A1732 was observed with the Australia Telescope Compact Array (ATCA) on 1995 April 18 at frequencies of 1.344 and 2.378 GHz. Total integration time was 10 hours in the 6C array, which gives interferometer spacings from 153 m-6 km. The synthesised half-power beamwidths at the declination of A1732 were $16.8'' \times 6.8''$ (PA $1.4^{\circ}$ ) at 1.344 GHz and $9.6'' \times 4.1''$ (PA $2.3^{\circ}$ ) at 2.378 GHz. The primary flux density calibrator was PKS B1934-638, with B1245-197 and B1622-297 as secondary phase calibrators. ISO J1324-2016 appeared as an unresolved radio source, with a fitted position of $13~24~45.67 \pm 0.03$ , $-20~16~11.3 \pm 0.5$ (J2000), and flux densities of $1.4 \pm 0.12$ and $0.8 \pm 0.1$ mJy at 1.344 and 2.378 GHz, respectively. The spectral index over this interval is $\alpha = -1.0$ ( $S_\nu \propto \nu^{\alpha}$ ). The 0.5'' radio positional accuracy was essential for the follow-up identification work.
Optical/NIR broad-band imaging
From 1993 CFHT images of A1732, the following upper limits on ISO J1324-2016 were set: B> 24.5, R > 22.7. However, since the source was close to the CCD edge, and affected by vignetting, these limits do not provide a useful constraint. The source was then observed in 1997 (March 05-06) at ESO with NTT/SUSI for two hours in the I band; the seeing was 0.6''. In the radio error box, we discovered an object with $I = 22.4\pm 0.1$ (see Fig. 1). The I image of the identification is unambiguously pointlike ( FWHM < 0.6''). Subsequently, we obtained a UKIRT service image (1998 April 21) of the field and measured a Kmagnitude of 17.5. This magnitude of $K = 17.5 \pm 0.1$ was later confirmed by the VLT/ISAAC acquisition image (see Fig. 1); moreover, with a seeing of 0.4'', the ISAAC K image of ISO J1324-2016 remains pointlike. The I-K color of 4.9 emphasises the extreme redness of the source spectrum - at least in the observed frame - and the present lower limit, R-K > 5.2, reinforces the status of ISO J1324-2016 as an ERO.

X-ray observations
The field of Abell 1732 was observed with the ROSAT HRI for $\sim$ 30 ks (1996 January 15-27; Pierre et al. 1996) and with ASCA for a total integration time of $\sim$ 90 ks and $\sim$ 100 ks by the SIS and GIS instruments respectively (1997 July 6-7; Pierre et al. 1999). ISO J1324-2016 is not detected by the HRI. This sets a 3 $\sigma$ flux upper limit of $\sim$ $2\times 10^{14}$ erg s^-1 cm^-2 in the [0.1-2.4] keV band, assuming a standard power-law spectrum with a photon index of 2 (or $4\times 10^{14}$ erg s^-1 cm^-2 if corrected for Galactic absorption). In the [0.4-10] keV ASCA images, because of the large instrumental point spread function, the cluster image encompasses the ISO J1324-2016 position, making its detection impossible.

The NIR spectroscopy
A 2-hour spectroscopic observation (1998 May 23) at intermediate resolution with EFOSC on the ESO 3.6-m telescope over the range 4000-9000 Å did not reveal any significant absorption/emission features. This stressed the need for deep NIR spectroscopy with a larger telescope. ISO J1324-2016 was observed with the low resolution spectroscopic mode of VLT1/ISAAC on 2000 June 7 & 9. Three grating settings were used to cover the 1.1-2.5 $\mu$ m range. As is standard in the IR, the target was observed at two positions along the slit (which we shall call the A and B beams). The bright and variable night sky lines were removed by subtracting the respective spectra from each other. The resulting two-dimensional spectra were then corrected for slit distortion and wavelength calibrated with the OH lines or with arc lamps. Residual lines from the night sky were then removed by combining spectra from the A and B beams. This process works well enough that one-dimensional spectra can be extracted without any need for additional sky subtraction. In addition to ISO J1324-2016, two hot stars, with spectral type A0 or earlier, were observed with the same instrument configuration. These stars were used to remove telluric features in the spectra of ISO J1324-2016. The one-dimensional spectra are shown in Fig. 2; spectral resolution is 21.4 and 28.4 Å in the J and H bands respectively. The redshift from the H $\alpha$ line is $1.500 \pm 0.002$ . The H $\beta$ and [OIII] lines show a relative blueshift of some 1500 and 800 km s^-1 respectively. The Balmer decrement is uncertain because the H $\beta$ line is very noisy (due to a sky line at 1.215 $\mu$ m) and appears to be narrower than H $\alpha$ ; the estimated value of 20⁺¹⁰_-4 is significantly higher than more common values of 5-10. The total corresponding B extinction would be A_B = 7⁺²_-1 in the source restframe, assuming standard extinction laws (Mathis 1990).

2 The data set