7 The IR variability

The total duration of our ISO campaign was 374 days, with a mean sampling interval of 24.9 days. In addition to the weighted-mean spectrum, $\langle F_\nu \rangle$, Fig. 3 displays the rms spectrum $\langle {\rm rms}_\nu \rangle$, and the mean error spectrum $\epsilon_\nu$. The latter was evaluated as

$$\epsilon^2_\nu = \frac{1}{N\,\sum_{i=1}^{N} 1/\Delta_{\nu,i}^2}$$ (1)

where $\Delta_{\nu,i}$ is the uncertainty associated with the flux F_i at epoch i and $N\,=\,16$ is the number of epochs. We can exclude significant (i.e. 3$\sigma$) flux variability since the largest value of $\langle {\rm rms}_\nu \rangle / \epsilon_\nu$ is only $\approx$ 1.4. A chi-square test was also applied to the average PHT-S fluxes of Table 1. The reduced chi-squares are <1 which confirms that no significant variations of the mid-IR flux took place during our $\sim1$-year observing campaign.