1 Introduction

IRAS scanned the north ecliptic polar region (NEPR) over 1000 times from all directions. Hacking (1987) coadded these scans, representing more than 20 hours of integration time, and obtained point source filtered maps used by Hacking & Houck (1987, henceforth HH87) to compile the deepest FIR samples available before the advent of ISO surveys. The $60\,\mu$m sample (henceforth referred to as the IRAS Deep Survey sample, IDS), probably constituted entirely of galaxies, has long been the most crucial piece of information on which numerous studies (see, e.g.: Ashby et al. 1996; Hacking et al. 1987; Oliver et al. 1992; Treyer & Silk 1993; Franceschini et al. 1994) of the far-IR evolution of galaxies relied. It is comprised of 98 sources (plus the planetary nebula NGC 6543) with $60\,\mu$m fluxes $\geq 50$ mJy in an area of 6.25 square degrees. Most (77) sources also have $100\,\mu$m fluxes (although 25 of these are rather uncertain); 17 were detected at $25\,\mu$m, but only 5 at $12\,\mu$m. Ashby et al. (1996) took optical spectra of 76 tentative IDS identifications at the Palomar 5 m telescope.

We have carried out ISO observations with the CAM LW3 filter (range $12-18\,\mu$m, $\lambda_{\rm eff} = 14.3\,\mu$m) of 94 IDS sources plus a source detected by IRAS at $25\,\mu$m and found to have particularly interesting properties.

The scientific rationale of these observations is severalfold. First, we aimed at assessing the reliability of faint $60\,\mu$m sources themselves and, hence, of their counts below 100 mJy. As pointed out by HH87, the flux uncertainties of their faintest sources is $\simeq 20\,$mJy so that a 50 mJy sources is a $2.5\sigma$ detection. At this low S/N level, the interpretation of the data requires a careful quantification of completeness, reliability and measurement biases. Spurious sources may be produced by the significant cirrus contamination affecting the field. Also, due to the limited angular resolution of the survey, some faint IRAS "sources'' may actually be multiple systems. In fact, there is a considerable uncertainty on the faint end of $60\,\mu$m counts, different studies yielding estimates differing by factors up to $\simeq 2$ (HH87, Gregorich et al. 1995; Bertin et al. 1997).

Second, ISOCAM observations allow a significant improvement of the positional accuracy of sources and a corresponding improvement of the reliability of optical identifications. An accurate location is essential since some of these sources may be optically faint either because of a high obscuration by dust or because they are very distant. But the large surface density of optically faint sources makes the identification process very uncertain unless the error box is correspondingly small. This is particularly critical to correctly trace cosmological evolution. For example, some models predict that a few IDS sources may be dust enshrouded galaxies at substantial redshifts; but these sources may easily be misidentified with brighter galaxies in the field.

Third, our measurements establish a direct link between IRAS and ISOCAM counts. The former cover fluxes up to two orders of magnitude brighter than the shallow surveys in the LW3 filter (Rowan-Robinson et al. 1999; Désert et al. 1999) and are therefore a very important complement to the latter.

Fourth, combining ISOCAM and IRAS data we can analyze the far-IR spectral energy distribution (SED) of a relatively large sample of far-IR selected galaxies, covering several Gyr in lookback time. It will be interesting to see, for example, if the cosmological evolution clearly indicated by the $60\,\mu$m counts is somehow reflected in an evolution of the SED. Fifth, from the bivariate $60\,\mu$m/$14.3\,\mu$m luminosity distribution and the $60\,\mu$m local luminosity function, we may derive an estimate of the poorly known $14.3\,\mu$m local luminosity function. In this paper, the first of a series, we present and briefly discuss ISOCAM data. In a second paper we will compare ISOCAM fluxes with IRAS and radio data and discuss the spectral energy distribution of sources in the far-IR to radio region. The third paper will deal with optical identifications, spectroscopy and photometry.