5 A weak magnetic field in $\eta$ Aql?

Although our results do not support the specific conclusions of Plachinda (2000), do they provide any evidence for the existence of weaker magnetic fields? The detection significance $z=\vert\ensuremath{B_{\rm z}} \vert/\sigma_{\rm B}$of our longitudinal field measurements ranges from 0 to 3.3, and the reduced $\chi^2$ of the sample as compared to a zero field model is 3.3. This value is somewhat larger than the maximum reduced $\chi^2$(2.6 at the 99.9% level) expected for 14 measurements, suggesting that a zero field model may not be an adequate description of the data. The reduced $\chi^2$ is however strongly influenced by two measurements ($-14 \pm 5$ and $-13 \pm 4$ G) obtained on JD 2452102 (excluding these two measurements, the reduced $\chi^2$ is just 2.3). It may be that these two measurements (and the general trend of negative longitudinal field after JD 2452101.608) reflect the presence of weak ($\sim$10 G) magnetic field (with characteristics apparently very different from the field claimed by Plachinda). However, we point out that the phase at which these two measurements were acquired (phase 0.92) is precisely the phase at which we expect the largest velocity fields in the atmosphere of $\eta$ Aql, due to passage of the primary shock wave. In fact, whereas we observe no detectable systematic change in radial velocity amongst the spectra obtained on 10 July (phase 0.759-0.787) or 12 July (phase 0.033-0.068), on 11 July (phase 0.915-0.928) we observe a clear systematic change of the RV of the mean line of about 4 km s$^{\rm -1}$. Although our observing and reduction procedure is designed to minimise spurious signatures due to stellar variability (among other causes), it is certainly clear that such rapid variability makes the danger of systematic errors much more severe. Therefore we note the apparent marginal detection of a weak magnetic field around phase 0.92, although we tentatively ascribe it to systematic error, pending further observations.