2 Observations and processing of the data

The experiment was launched on February 7, 2002 by the CNES from the Swedish balloon base in Esrange, near Kiruna, Sweden, $68^\circ$N, $20^\circ$E. It reached a float altitude of $\sim$34 km and landed 21.5 hours later in Siberia near Noril'sk, where it was recovered by a Franco-Russian team. The night-time scientific observations span 11.7 hours of integration from 15.3 UT to 3.0 UT the next day. Figure 1 shows the Northern galactic part of the sky observed during the flight.

Figure 1: Archeops CMB map (Galactic coordinates, centered on the Galactic anticenter, Northern hemisphere) in HEALPIX pixelisation (Gorski et al. 1998) with 15 arcmin pixels and a 15 arcmin Gaussian smoothing. The map is a two-photometers coaddition. The dark blue region is not included in the present analysis because of possible contamination by dust. The colors in the map range from -500 to  $500~\mu{\rm K_{CMB}}$.

A detailed description of the data processing pipeline will be given in Benoît et al. (2003c). Pointing reconstruction, good to 1 arcmin, is performed using data from a bore-sight mounted optical star sensor aligned to each photometer using Jupiter observations. The raw Time Ordered Information (TOI), sampled at 153 Hz, are preprocessed to account for the readout electronics and response variations. Corrupted data (including glitches), representing less than 1.5%, are flagged. Low frequency drifts correlated to various templates (altitude, attitude, temperatures, CMB dipole) are removed from the data. To remove residual dust and atmospheric signal, the data are decorrelated with the high frequency photometers and a synthetic dust timeline (Schlegel et al. 1998).

The CMB dipole is the prime calibrator of the instrument. The absolute calibration error against the dipole measured by COBE/DMR (Fixsen et al. 1994) is estimated to be less than 4% (resp. 8%) in temperature at 143 GHz (resp. 217 GHz). Two other independent calibration methods, both with intrinsic uncertainty of $\sim$10%, give responsivities relative to the dipole calibration at 143 (resp. 217 GHz) of -5 (resp. +6%) on Jupiter and -20 (resp. -5%) with COBE-FIRAS Galactic Plane emission.

Figure 2: The Archeops CMB power spectrum for the combination of the two photometers. Green and red data points correspond to two overlapping binnings and are therefore not independent. The light open diamonds show the null test resulting from the self difference ( SD) of both photometers and the light open triangles correspond to the difference (D) of both photometers (shifted by $-2500~\mu{\rm K}^2$for clarity) as described in Sect. 4 and shown in Table 1.

Figure 3: Contamination by systematics: the Archeops CMB power spectrum statistical error bars (including noise and sample variance) are shown as the blue triangles. The large error bar in the first bin mainly comes from the high-pass filtering. A conservative upper-limit to contamination by dust and atmospheric signal is shown in red crosses, with a $\ell$ different binning to enhance the low $\ell$ side. Beam and time constants uncertainties are shown in dot-dashed blue and dashed green (see text). The 7% temperature calibration uncertainty is not shown here. The window functions are shown at the bottom of the figure.

The beam shapes of the photometers measured on Jupiter are moderately elliptical, having a ratio of the major to minor axis of 1.2 (resp. 1.5) at 143 GHz (resp. 217 GHz), and have an equivalent FWHM of 11 arcmin (resp. 13). The error in beam size is less than 10%. The effective beam transfer function for each photometer, determined with simulations, is taken into account in the analysis and is in excellent agreement with analytical estimates (Fosalba et al. 2002).

Figure 4: The Archeops power spectrum compared with results of COBE, Boomerang, Dasi, Maxima (Tegmark 1996; Netterfield et al. 2002; Lee et al. 2001; Halverson et al. 2002).