Tadpole galaxies can be lopsided and still orderly inside
MUSE spectroscopy of twenty tadpole galaxies in the Hubble Ultra Deep Field finds their gas moving along their tails, at a median 12° between the shape's axis and the motion's axis where unrelated axes would average 45°.
Twenty tadpole galaxies in the Hubble Ultra Deep Field, in an arXiv preprint, have their gas moving along the length of their tails. The median angle between the long axis of a galaxy's shape and the axis of its gas motion is 12°. Two axes with no relationship to each other average 45°. That misalignment is the number the argument turns on: a lopsided outline does not require a chaotic interior. Thirteen of the twenty show a clean velocity gradient across the gas as well, and the seven that do not are a puzzle of their own.
Tadpole galaxies are named for their outline: a compact bright head with a tail of fainter material drawn out behind it. The shape is solid evidence that something disturbed the galaxy. It is no evidence of what disturbed it, or whether that disturbance is still running. A lopsided outline does not tell you which way the gas is moving.
Where the shape comes from, and where the motion comes from
The two measurements at the centre of the paper are made on two different substances, and that is the point of it.
The shape was measured from imaging, with Hubble and JWST recording where the light is. Almost all of that light is starlight, and all of the tail is: the tail is a stream of stars pulled out of the galaxy and left with nowhere to put the energy of the encounter.
The motion came from MUSE, a spectrograph on the Very Large Telescope that takes a spectrum in every pixel across a small patch of sky instead of one spectrum down a single slit. Pointed at a galaxy, it returns thousands of spectra at once, and each one reduces to a single number: how fast the gas in that pixel is moving toward or away from us. MUSE sees ionised gas. The paper's causal claim rests on gas and stars being two different substances that forget a disturbance on different timescales.
The line MUSE measured is the [O II] doublet, two emission lines from singly ionised oxygen at rest wavelengths of 3726 and 3729 ångströms. The square brackets are the standard notation for a forbidden line, one that an atom can only emit in the sparse, low-density gas between the stars. The two components sit under three ångströms apart, close enough that in a distant galaxy they arrive as one blended feature.
What makes them useful is the Doppler shift. Gas moving toward the telescope has its wavelength compressed toward the blue; gas moving away has it stretched toward the red. Fit the centre of the line in each pixel and the velocity map falls out directly: the galaxy's internal motion, laid out spatially rather than averaged into a single spectrum. Across the sample's stated redshift range, 0.4 to 1.5, those two rest wavelengths work out to about 5200 to 9300 ångströms as observed.
What the twenty maps contain
The classification is a judgement, not a hard cut. What is being tested is whether a galaxy-scale velocity gradient is present: a systematic change in the [O II] velocity from one side of the galaxy to the other. The abstract's own phrase for the thirteen is "moderate velocity gradients", and it quotes no threshold in velocity or in significance. So ordered here means a gradient was detected across the galaxy's face. It does not mean a rotation curve was fit and passed a test.
| Measured | If a lopsided outline meant a disturbed interior | 20 tadpole galaxies, z = 0.4 to 1.5 |
|---|---|---|
| Ordered 2D velocity field | rare | 13 of 20 |
| Angle between the shape's long axis and the motion's axis | near 45°, the value for unrelated axes | median 12° |
| Offset between the [O II] centroid and the Hubble/F775W centroid | kpc-scale | below 2 kpc in 16 of 20 |
Two randomly oriented directions differ by 45° on average, because every angle from 0° to 90° is equally likely. The tadpoles' shapes and their motions differ by a median 12°, so the gas is very often moving along the length of the galaxy rather than across it.
The third row is an upper limit on displacement, not a measurement of it. Two kiloparsecs is roughly a quarter of the way from the Sun to the centre of the Milky Way. Gas and stars respond to a disturbance over different timescales, which is why their centroids are worth comparing; but two very different situations leave the centroids close together, and the paper does not say which of them it is looking at. A galaxy early in a disturbance, where the gas has barely begun to decouple, and a galaxy whose gas has already found its way back to the stellar centre leave the same signature. In 16 of these 20 galaxies the two centroids have not separated by more than 2 kpc.
The part that stays open
An ordered velocity field is not the same thing as a rotating disc, and the abstract makes no claim of rotation. It describes evidence for ordered motion and moderate velocity gradients.
What one emission line cannot do is separate the cases. A single velocity gradient is also what a bipolar outflow produces, with gas streaming from both poles along one axis, and what a pair of galaxies falling toward each other along the line of sight produces. Two-dimensional maps built from one line cannot tell them apart on their own. That is a general caveat rather than one the paper raises, so the honest reading of the thirteen is a gas disc that is coherent across the galaxy, not one that has been diagnosed. Whether the flow is rotation or a passing outflow is what the next measurement has to decide.
The sample is small and shaped by the question. Twenty galaxies, all in the Hubble Ultra Deep Field, a single small patch of sky, selected in the first place because they looked like tadpoles. The seven with no ordered motion fit the older picture perfectly and get one line in the abstract. Whether those seven really are disturbed, or whether their [O II] emission is simply too faint and patchy to fit a gradient to, is not settled here.
Why the shape stops being a shortcut
Astronomers sort galaxies by shape. It is fast, it works on objects far too faint for spectroscopy, and it has organised the field since Hubble's tuning fork. The assumption underneath the sorting is that a galaxy that looks a certain way moves in a certain way. The counterexample is already in hand: thirteen galaxies with lopsided outlines whose gas runs along one axis, at a median 12° to the long axis of the shape.
The paper's claim about why is narrow. It says that morphological and kinematic settlement "does not occur simultaneously", and that these galaxies "might represent a transient evolutionary phase". The physical reason to expect that gap is that gas and stars shed a disturbance on different terms. Ionised gas is dissipative: a shock converts ordered motion into heat, the heat radiates away as light, and the gas can come back to a coherent flow. Stars are effectively collisionless. There is nothing for them to radiate the energy of the encounter into, so they hold the memory of it in their orbits, and a tail disperses only over the much longer dynamical time it takes to wrap up and mix into the rest of the galaxy. The gas has a channel to shed the disturbance. The outline does not. That reasoning is the interpretation these twenty galaxies are consistent with, not something they measure.
What would settle it is a sample selected without a shape cut, so the tadpole fraction falls out of the data instead of being imposed on it, and kinematics measured in the stars rather than in the ionised gas. The paper proposes neither. Stars cannot radiate a disturbance away, so they hold onto it long after the gas has come back to order. If the stars in these thirteen galaxies are still disordered while the gas is coherent, the transient-phase reading holds. If the stars are ordered too, the tails are older than they look, and the disturbance happened further back than the ionised gas can now show.