We measure the angular power spectrum of the WMAP first-year temperature anisotropy maps .
We use SpICE ( Spatially Inhomogeneous Correlation Estimator ) to estimate C _ { \ell } ’ s for multipoles \ell = 2 - 900 
from all possible cross-correlation channels .
Except for the map-making stage , our measurements provide an independent analysis of that by ( ) .
Despite the different methods used , there is virtually no difference between the two measurements for \ell \lower 2.15 pt \hbox { $ \buildrel < \over { \sim } $ } 700 ; the highest \ell ’ s are still compatible within 1 - \sigma errors .
We use a novel intra-bin variance method to constrain C _ { \ell } errors in a model independent way .
Simulations show that our implementation of the technique is unbiased within 1 % for \ell \lower 2.15 pt \hbox { $ \buildrel > \over { \sim } $ } 100 .
When applied to WMAP data , the intra-bin variance estimator yields diagonal errors \sim 10 \% larger than those reported by the WMAP team for 100 < \ell < 450 .
This translates into a 2.4 \sigma detection of systematics since no difference is expected between the SpICE and the WMAP team estimator window functions in this multipole range .
With our measurement of the C _ { \ell } ’ s and errors , we get \chi ^ { 2 } / d . o . f . = 1.042 for a best-fit \Lambda CDM model , which has a 14 % probability , whereas the WMAP team ( ) obtained \chi ^ { 2 } / d . o . f . = 1.066 , which has a 5 % probability .
We assess the impact of our results on cosmological parameters using Markov Chain Monte Carlo simulations .
From WMAP data alone , assuming spatially flat power law \Lambda CDM models , we obtain the reionization optical depth \tau = 0.145 \pm 0.067 , spectral index n _ { s } = 0.99 \pm 0.04 , Hubble constant h = 0.67 \pm 0.05 , baryon density \Omega _ { b } h ^ { 2 } = 0.0218 \pm 0.0014 , cold dark matter density \Omega _ { cdm } h ^ { 2 } = 0.122 \pm 0.018 , and \sigma _ { 8 } = 0.92 \pm 0.12 , consistent with a reionization redshift z _ { re } = 16 \pm 5 ( 68 \% CL ) .