One of our arms is missing, send help

More follow-up from last week's low surface brightness conference. Actually this paper came out just before, but I didn't have time to do a write-up. And as it happens that's probably for the best.

Galaxies aren't very similar to cats or tables or bananas, though I expect you knew that already. You probably also know that those objects do look a bit different depending on what wavelength you use - a banana, for instance, doesn't look that much different from the background to a thermal camera, whereas a cat is much more obvious. But in all cases, the basic shape doesn't really change no matter what you do*. A cat doesn't start looking like a house if you take a long enough exposure. A table isn't revealed to have very faint but enormous tentacles by looking in the infra-red.

*Unless you go to really extreme values, where resolution and sensitivity effects dominate. E.g. a 1m wavelength radio wave wouldn't be a good way to measure a 0.5m cat.

For galaxies it's a different story altogether. The more thinly spread the stars, the harder they are to detect (and the same goes for the gas). And galaxies are pretty robust things but they're not invulnerable - interactions with other galaxies can distort them or tear bits off, often causing long, very very faint stellar streams. Although such features tend to only comprise a tiny fraction of the galaxy's total stellar mass, they can give important clues as to what's happened to that galaxy in the past. Which means that much of the really exciting stuff that the galaxy's done can only be revealed by taking really deep imaging.

Take, for example, the edge-on galaxy NGC 5907. In a short exposure it looks a bit like this :

Which is very nice but entirely normal. But in a much deeper image you see this :

There are various reasons the colours are different but they're not important here.

Much more interesting ! Except the authors of the study below have done their own deep imaging, and they find something different again :

This is from the same team who led the charge to the current renaissance in low surface brightness galaxies that's got everyone interested again. It's a rather easier feature to intuitively understand, at least considering only the main, brightest parts - it certainly looks like what you'd expect from a tidally-disrupted dwarf galaxy. The authors show the results of a simple simulation that reproduces this result very well, and even identify the possible main remnant of the progenitor satellite.

Except... why does it look so darn different from the previous image. Here's where they come unstuck, which needs to be quoted at length :

We now turn to the most puzzling aspect of our study. The morphology of the stream in our data is a good match to the shallower discovery image of Shang et al. (1998), and also to a meta-analysis of NGC 5907 images by Lang, Hogg, & Scholkopf (2014) and to Subaru imaging (see Laine et al. 2016, and S. Laine, priv. communication). However, it is qualitatively different from that reported by M08. The most striking difference is that we do not confirm the presence of the second loop...  

We cannot definitively determine the cause of these discrepancies with M08, but a likely explanation lies in the image processing procedures that were applied to the data. The M08 data were obtained and processed by an experienced amateur astronomer. Amateurs have played an important role in this field as they convincingly demonstrated the power of small telescopes for low surface brightness imaging (see Martínez Delgado et al. 2010). However, the methods that are used by the amateur community typically do not allow for quantitative analysis, as their image processing is generally optimized for aesthetic qualities rather than preserving the linearity and noise properties of the data.

And that's it. But this just isn't good enough. It's fine to quote the sources which also don't show this second loop, but at the conference we were shown a whole panel of independent images which did also show the second loop. And processing for "aesthetic qualities" does not mean "photoshopped to look nice" - some amateurs are really bridging the gap between professionals and hobbyists. It seems fantastically unlikely to me that that the second loop, a feature which is so clear and distinct in the middle image above, is some kind of artifact. You'd need one heck of a bizarre processing job to get something so coherent if it wasn't real, especially one that's been re-detected observationally by different observers with different telescopes.

My impression is that this paper should have been a full paper, not a letter. Sure, it's a newsworthy and eye-catching result. But with all that work gone into it, why not write a more detailed report ? There's no reason to rush that I can see. I had the feeling that this result has only made the community more skeptical of the authors' growing number of other controversial results. On a more positive note, interest in the faintest features is definitely growing, so expect more such dramatic discoveries and similar controversies ahead.

Dragonfly imaging of the galaxy NGC5907: a revised view of the iconic stellar stream

In 2008 it was reported that the stellar stream of the edge-on spiral NGC5907 loops twice around the galaxy, enveloping it in a giant corkscrew-like structure. Here we present imaging of this iconic object with the Dragonfly Telephoto Array, reaching $1σ$ surface brightness levels of $μ_g\approx 30.5$ mag/arcsec$^2$ on arcminute scales.