How big is an arc second anyway?

It's a very small angle, but how does it compare to a second of time? Can FirstScope see an arcsecond? Can we detect the changes in the speed of rotation of the Earth which lead to leap seconds?
2026-10-30 · firstscope v1

Seconds of arc and seconds of time

Where the stars are in the sky is governed by the rotation of the Earth, and since that rotation varies compared with our modern clocks, it has to be taken into account. Like the arc minute, the arc second is an unfortunate term, because here it sits next to the second of time: if the Earth is one second late in its rotation, which is how large these variations can add up to over a few years, the stars will be displaced to the east by 15 arcseconds.

A FirstScope pixel (strictly a photosite: one red, green or blue site) is about 1 arcsecond across, 1.07 with the HQ camera, so the stars cross it in 71 milliseconds of the Earth's turning. That is at the celestial equator, though, and the FirstScope looks at the celestial pole, where the sky barely moves. Polaris, two-thirds of a degree from the pole, covers only 0.17 arcseconds a second, so there one photosite is 6 seconds of the Earth's turning.

The Earth is late

Just now the Earth is late. It does not turn at a steady rate, and our clocks, which run on atoms, take no notice. The gap has a name, DUT1: the time told by the Earth's turning (UT1) minus the time on the clock (UTC).

Being negative, it means every star arrives late: tonight a star crosses the meridian 53 milliseconds after the clock says it should, and is that much east of its predicted place. Those figures are for the celestial equator. Away from it they shrink like right ascension does, with the cosine of the declination.

The ground is what is out of place, of course, not the stars. Here in London we travel at 290 metres per second, so the house is about 15 metres short of where an Earth keeping clock time would have put it.

If the gap ever nears 0.9 seconds, 13.5 arcseconds, UTC gets a leap second to close it. The last was at the end of 2016.

The ladder

Carrying on down from How big is an arc minute anyway?:

The other third of an arcsecond

The Earth also wobbles on its axis, so the pole wanders over the ground. Today it is 0.32 arcseconds one way and 0.14 the other from the reference position: 0.35 arcseconds, or 11 metres on the ice. That moves the sky over London by about 0.3 arcseconds. It is the same size as DUT1 was a month ago, and it wanders in loops rather than growing.

Can the FirstScope see it?

Just about, in one sense, and not in the sense that matters.

It cannot resolve it. 0.8 arcseconds is under one photosite and less than half the blur of the mirror. At the focus a third of an arcsecond is half a micron, which is about one wavelength of green light.

It can locate it. Resolving is telling two things apart. Locating one star is much easier: the centre of a blur can be measured to a small part of its width, about a hundredth of an arcsecond for a well-focused star at 100 times the noise. So the optics and the sensor have what it takes, once the focus does. They have not shown it yet. The best measured star so far is 11.5 arcseconds wide (September, an upper bound), and with the HQ camera the scope is still 2.7 mm of travel short of focus on the stars.

The trouble is that every star moves together. DUT1 shifts the whole sky as one, so no picture shows it. A plate solve would just report that the scope points a little to the east. To see it, the scope must be a fixed reference: time stars across the same pixel on a run of nights, and watch them come 15 thousandths of an arcsecond later each day. That is how the Earth's turning was measured before atomic clocks. It asks a lot:

So the FirstScope's numbers reach this scale and the table it stands on does not.

For the other cameras it is not even a question. DUT1 today is about a seventy-fifth of an astrocam v3 photosite and a hundred-and-thirty-fifth of an eclipticam v3w one at the centre of its frame. On the FirstScope's September field, 4 degrees from the pole, it would be just 0.06 arcseconds.

scaleunitsarcsecondsdut1earth-rotationpolar-motionfirstscope
← When does last night arrive? It depends on the SunHow big is an arc minute anyway? →