In a borderline eclipse of the Moon like last Saturday's, the difference between "total" and "partial" depends on some crucial assumptions. Here's the whole story at last.
An especially intriguing eclipse of the Moon delighted observers from western North America to Australia and the Far East last Saturday (April 4, 2015). What still makes it so intriguing, several days later, is the question of whether the world saw a barely total lunar eclipse — the third in a "tetrad" of four, as almost universally predicted — or a barely partial lunar eclipse, with a hair-thin rim of sunlit lunar surface remaining just outside the umbra of Earth's shadow. Which would spoil the tetrad.
Visual observers are saying it was barely partial.
You'd think astronomers would be awfully embarrassed not to get such a simple prediction right. But the edge of Earth's umbra is fuzzy enough, and the umbra just inside the edge is bright enough — and the effects of Earth's atmosphere on the umbra's precise size are unpredictable enough — that we may never be sure which kind of eclipse we saw on Saturday.
The evidence certainly leans toward partial in terms of how people judge what they saw. But the most detailed mathematical analysis says it was theoretically total by a hair... probably.
Views of the Lunar Eclipse
"I observed the eclipse from San Francisco, naked eye and through mounted 11×56 binoculars," writes Anthony Barreiro in a typical report. "At no time did the eclipse appear total to me. The Moon’s far northern limb remained brightly illuminated throughout. Inside the apparent edge of the umbra the Moon’s surface was bright but obviously reddened. The very edge of the Moon remained much brighter and without any visible reddening."
Scott Bulkley, who has observed lunar eclipses for more than 50 years, used naked-eye observations, 10×50 binoculars, and a 5-inch Schmidt-Cassegrain telescope for this one. "Using all three viewing methods during the critical minutes, I was not able to observe totality at any time," he writes. "There was always at least a very thin ‘sliver’ of brightly illuminated lunar surface on the northern limb of the Moon."
Alastair from New Zealand writes, "I am sure this was not a fully total eclipse. A thin sliver of the limb was still quite brightly illuminated throughout totality."
All the other reports we’ve received so far agree.
But the problem with such reports is that the innermost penumbra — the portion of Earth's shadow just outside the umbra — was not necessarily in view, so there was nothing that the relatively bright umbral rim could definitely be compared against.
Nor does imaging help. Imaging is actually worse for determining the umbra's edge, because the amount of exposure, contrast, dynamic range and so forth can vary far more between different images than between different people's eyes at a telescope. Not to mention the tweaking and processing that may happen automatically in the camera, or by the photographer to "correct" the image later.
Eclipse Theory and Eclipse Observation
Wouldn't the simple geometry of the Earth, Moon, and Sun — known with extreme precision — give an exact answer?
The problem here is that Earth's atmosphere enlarges Earth's size, for shadow-casting purposes, by a not-quite-exact amount. The U.S. Naval Observatory, in its lunar-eclipse predictions, adds a standard 2% to the umbra's "geometric" radius to account for the effect of the atmosphere. This led the USNO to predict 12.2 minutes of totality. The French national almanac office, and eclipse predictor Fred Espenak, used the supposedly more sophisticated "Danjon method." This gave 4.7 minutes of totality and a much thinner margin for error.
However, in the days before the eclipse, Chicago amateur Curt Renz pointed out that neither method includes Earth's oblateness. Our planet is 1⁄300 less wide from pole to pole than it is across the equator (due to rotation). And Earth's far northern latitudes were the parts that cast the shadow's eclipse-defining northern edge.
Normally Earth's oblateness is insignificant. But in such a borderline case, might it tip the eclipse into being partial?
David Herald has carried this possibility through a full analysis, including both the oblateness and the best new information about the size of the umbra from a massive study of lunar eclipse crater timings recently done by himself and Roger Sinnott. Herald writes,
The most recent approach (Herald & Sinnott, JBAA 124-5 pp. 247-253, 2014) is based on the Danjon approach; however it treats the Earth as oblate, allows for the varying inclination of the Earth relative to the Sun during the year, and increases the Earth’s [shadow-forming] radius by 87 km — that being the best fit to 22,539 observations made between 1842 and 2011. For this eclipse the magnitude is computed as 1.002.
In other words, total by just one part in 500 in term's of the umbra's radius! "However," Herald continues,
the increase in the Earth’s [apparent] radius required to fit the [historical] observations is not a straightforward issue. When you analyse the distribution of observations, you find there is a significant difference between the modal (78 km), median (85 km) or mean (86.9 km) values of the distribution. We adopted the mean value.
With a magnitude of 1.002, the theoretical edge of the umbra was beyond the limb of the Moon by a mere 4.6 arcseconds at maximum eclipse. This is quite small compared to the extent of the transition at the edge of the umbra.
The magnitude of this latest eclipse, computed using the various values, comes out as:
Modal (78 km): mag = 0.999
Median (85 km): mag = 1.001
Mean (86.9): mag = 1.002
Unless you have a clear understanding of the nature of the statistical variation, and the basis on which you are computing the magnitude of the eclipse, you may well be misled into the conclusions to be drawn from a small sample of observations. Given a statistically large number of observers, for this eclipse one would expect that the ‘most common’ estimate of the magnitude would be 0.999, that there would be as many estimates smaller than 1.001 as those larger than 1.001, and that the overall average of all estimates would be 1.002.
In short – those expecting a single well-defined value are doomed to disappointment.
The umbra does at least have an edge that's precisely definable, regardless of how the eye and brain might judge it. That edge is the place in the fuzziness where the change in brightness from point to point is steepest. This is the most natural definition of the umbra's edge, and the one that visual observers generally try to use when timing when the edge crosses craters and other lunar markings. These timings provide the best fix we have on what the umbra actually does.
But there's more uncertainty too! Sinnott and Herald also found that from one eclipse to the next, the atmosphere's effective eclipsing layer can differ in thickness by a few kilometers for reasons unknown. Unpredictably. They describe their work in the forthcoming June issue of Sky & Telescope.
We don't yet know how many people may have done good crater timings for this particular eclipse.
So, was it total or partial?
Visual observers seem to agree that it appeared partial. But by the exact definition of the umbra's edge? We'll probably never know.
For a closer look at our nearest celestial neighbor, check out our Moon globe, pieced together from thousands of Lunar Reconnaissance Orbiter images.
And to prepare for the last of the tetrad on September 27th, be sure to download our free lunar eclipse ebook to get a play-by-play of the celestial event plus a free Moon map.