Yes, and they say the interp is subject to detailed modelling....

Yes, and they say the interp is subject to detailed modelling. If you enlarge the plots you can see the data is noisy, with what looks to be considerable variance between repeat readings of the same channels (IMO). As well as an overall untidy response in the z component (last plot). But you couldn't seriously say it is the lost rods in the other hole as they would be difficult (probably impossible) to energise. Either way I wouldn't be twisting any muselets (word of the day) from atop champagne bottles just yet.