Canaletto and the Moon: Part 1- La Vigilia di San Pietro
It all started in Berlin. On a chilly, overcast day after Christmas 2017, Alice and I flew to Germany for a short break. Our plan was to visit the Gemäldegalerie to seek out some paintings by some of Alice’s favourite artists — Velázquez, Rembrandt, van Dyck.
At some point in our self-guided meander through the collection, I found myself standing in front of a series of four large Venetian views by Canaletto. Hung from left to right along a single wall, they were (i) Grand Canal looking South-East from the Palazzo Michiel dalle Colonne to the Fondaco dei Tedeschi, (ii) Campo di Rialto (iii) La Vigilia di Santa Marta and (iv) La Vigilia di San Pietro.
I’ll admit that Canaletto was not an artist I’d paid any great attention to previously, but a couple of aspects of these paintings gave me pause. Of the four, the two ‘vigils’ are night scenes — moonlit with deep blue skies — most unusual for Canaletto.
Berlin’s Gemäldegalerie is an institution that lets its collection speak for itself. No context or background text explains the paintings to the viewer. Like all the objects in the gallery, the first night scene was simply labelled: a title, the artist’s name (and nickname) and years, and some approximate dates when it is believed to have been painted: 1758–1763. Other than a note indicating the paintings were on loan, that was it.
La Vigilia di S. Pietro
If you were in Venice some 250 years ago or more, this evening — June 28 — might well have been marked in your calendar. It is one of the famous Venetian festivals that are scattered through the year, not a formal celebration of the glories of the republic but, instead, a folk festival enjoyed by the people of the city. The feast of San Pietro is celebrated on June 29, and the festa or vigilia was celebrated the preceding evening, continuing well into the night.
The Festa di San Pietro is still celebrated to this day in towns across Italy, including Venice (although sadly, of course, not this year). While the celebrations don’t quite appear the same as in mid-eighteenth century, the geography of this part of the city is little changed. You would easily recognise the view:
The viewpoint depicted in Canaletto’s painting is no longer reachable on foot, having been walled in by the expanded Arsenale. We can see how it formerly was arranged in Lodovico Ughi’s famous 1729 map of Venice:
The painting’s viewpoint is marked at the red star. The campanile (bell tower) is indicated by the green star, and the approximate field of view of the painting is indicated by the triangle.
An astronomical clue?
Information overload is common in today’s world. With so much knowledge captured, documented and accessible via the internet, it’s easy to assume all the answers are already out there. The terse information presented in the Gemäldegalerie seems in marked contrast to the prevailing trend, but it has the advantage of allowing the curious visitor to pose their own questions.
The uncertain dating of the paintings floated across my mind: 1758–1763, a six-year window (all four paintings were similarly dated). The two night scenes each show the moon in a distinct phase, in identifiable locations, depicting the night before a saint’s feast day — well established dates in the calendar. The second night scene depicts La Vigilia di S. Marta celebrated a month later on July 28 — more on that in a later post.
Having spent the past several years working on The Photographer’s Ephemeris (TPE), an app that calculates the position of the sun and moon, a question formed: could the moon provide any clues to help date the paintings?
Cutting to the chase: no. I don’t think it can.
But the journey to research the answer has been a fascinating one, if art historical and astronomical detective work is your thing! Here’s some of what I learned about San Pietro.
Phase of the moon
The moon is shown neither full nor new, but somewhere in between. Can we be more precise?
As Renzo Leonardi notes[i], in Italy it is said, “gobba a levante, luna calante; gobba a ponente, luna crescente”, that is, if the darker part of the moon lies on the left side for the observer, the moon is waxing, and if on the right side, it is waning.
Canaletto’s moon is waning. He appears to have made some effort to indicate the lunar maria, but overall, the moon appears to be more than 50% illuminated, i.e. we have a waning gibbous moon.
Position of the moon
The moon is depicted in the southeastern sky. If we take Canaletto’s perspective at face value, then the approximate azimuth appears to be ~156°, measured by placing the grey map pin on the centreline of the rightmost building in TPE:
However, we must be careful: as David Bomford notes[ii], “there are many cases in which [Canaletto’s] departure from reality was extreme”. Canaletto’s use of a ‘synthetic’ viewpoint is a recurrent theme throughout his career. Another of the four Berlin Streit paintings, Campo di Rialto, is a case in point — the painting shows a synthesis of three or more distinct “on-the-ground” viewpoints.
But the painting and the actual topography are in good enough agreement that we can safely assume a south-southeast direction for the moon.
What about the altitude above the horizon? There are a couple of ways to go at it.
First, assuming the moon is painted to scale — we’ll return to that point — we can take some basic measurements from the painting to estimate the altitude to around +11 degrees:
The logic here is that the moon’s angular diameter is around half a degree. It occupies around 33 pixels on the screen (see below) and the upper limb lies 721 pixels above the estimated horizon line, implying it is around “21 moons” above the horizon, which equates to ~11°:
However, a more reliable approach is to use the height of the campanile. We can see that the moon is aligned close to the top of the campanile. On our visit to the Castello sestiere in Venice last year we took some measurements, using the app Theodolite:
Standing at the west end of the footbridge, I measured the very top of the campanile to be at +27.7. The distance measured on the map using TPE from that point to the base of the bell tower is 84m. The distance from Canaletto’s viewpoint is around 93m. Using some simple trigonometry, we can calculate that the angle to the top of the campanile in the painting (minus the dome, which appears to have been lost) is ~25°. The moon sits just below that point, so let’s say ~+24°.
That’s rather more than +11°. The difference illustrates the dangers in this sort of approach.
If +11° is correct, then Canaletto has shrunk the campanile significantly. Given his habit of flattering the famous campanile of the Piazza San Marco by elongating its proportions, it’s unlikely he would choose to shrink this one.
If +24° is correct, then the moon is painted at more than twice its actual size — a far more likely scenario. If you’ve ever photographed the moon, it’s common experience that it disappears to almost nothing in a wide-angle view. Painters have long exaggerated the size of the sun and moon to meet the expectations of our mind’s eye (e.g. Ippolito Caffi, in his depiction of the 1842 solar eclipse in Venice, paints them at eight times the actual size).
One more consideration: Canaletto’s viewpoint appears elevated above the ground — the horizon line seems to fall above head height of the figures in front of the campanile. It’s the classic primo piano (‘first floor’) view. The effective angle to the top of the campanile is therefore somewhat reduced.
Let’s go with a range of +15° to +25° for the target altitude.
What if we find a match?
We have the location of the viewpoint, and the azimuth, altitude, and phase of the moon. Together, they are sufficient to see if conditions on the ground on the night of Jun 28/29 are a match for the painting during our six-year period.
Before we do that, let’s ask, what if there were to be a good match?
It might be considered reasonable supporting evidence that the view Canaletto depicts represents what he personally observed. But there are several practical objections we’d need to overcome in order to believe this could be used to support dating the painting.
First and foremost, he didn’t paint it on the night. He didn’t typically paint en plein air at all. He sketched on location and then returned to his studio to work up his canvasses. Painting in oil on a large canvas in the dark, in the middle of a street party, is an unlikely scenario indeed.
We would have to believe that he recorded the position of the moon or remembered it well enough to reproduce it in the finished work after the event. Or perhaps he returned a few days afterwards to record the scene on a quieter evening. Why believe it is necessarily the moon of June 28 that is shown? Why should it be a particular moon at all?
A painting of a night scene requires a light source somewhere in the composition to illuminate the subject. The light of the moon is an obvious way to achieve this, particularly to illuminate large buildings beyond the reach of the small lamps and fires that dot the scene.
Would not Canaletto have painted the moon to function well artistically? Why should he have felt obliged to paint exactly what he had observed?
Imagine if the moon that night had been hanging in the sky in the centre of the view. The scene would have been evenly lit, with the north side of the righthand buildings in light — less dramatically directional, for sure. What if it had been in the sky to the left-hand side of the painting? The south-facing aspect of the Basilica, oriented to the viewer, would have been left in shadow. What if there was no moon at all? Would he have omitted it?
The moon located where it is, sitting above the right-hand building group, is arguably the strongest position both compositionally and in terms of directional lighting. It seems entirely plausible that this was an artistic choice, rather than merely following the lunar “facts on the ground”.
Is there a match?
Can we find a year where the moon conditions match those depicted? Here’s what I found:
- 1758 — no match: the moon is set, although the phase is a reasonable match
- 1759 — no match: the moon is setting behind the viewer; waxing crescent
- 1760 — partial match: azimuth and altitude are close, but it’s a full moon
- 1761 — no match: the moon is set; waning crescent
- 1762 — no match: the moon is out of view to the right; 1st quarter (waxing)
- 1763 — match: if you’re OK with a late night out (~1:30am), this matches azimuth, altitude and phase (waning gibbous)
The links above are set to 10pm Rome time (which is pretty much the same as Venice local mean time), with the exception of 1763, which I adjusted to the early hours of Jun 29 to achieve a match on the azimuth — the precise time doesn’t matter so much. This was a multi-hour event.
So, do we have an argument in favour of 1763? On the face of it, yes.
Unfortunately, this is where astronomical hope is dashed on the rocks of contemporary documentary evidence. Sigismund Streit, the retired German merchant who commissioned the four works from Canaletto, records that by May 26, 1763 all four paintings had been sent to Berlin[iii].
Could it be 1757?
In autumn 2018, I stumbled over a recently published book titled “Further Adventures of the Celestial Sleuth” by Donald W. Olson[iv]. Olson applies his skills as an astronomer to explaining various cultural depictions of celestial events in paintings, literature and more. Lo and behold, he also has looked into Canaletto’s night scenes.
Olson identifies the year the work was painted as 1757, having examined the years 1753 to 1763. This is based on searching for a waxing gibbous moon in the southeast sky in the evening of June 28.
Certainly, the time, azimuth and altitude appear plausible (TPE), but I can’t reconcile the moon in the painting with Olson’s identification as a waxing gibbous moon. As viewed from mid-latitudes in the northern hemisphere, waxing gibbous moons are darkened on the left side from the observer’s viewpoint, not the right, as shown in the painting: “gobba a levante, luna calante”.
Olson also identifies 1757 as the year for La Vigilia di Santa Marta, implying the two paintings were based on festivals observed the same summer.
However, this raises a further issue. You’ll recall that the Gemäldegalerie dates the paintings to 1758–1763. Why those years? 1763 is established by the documentary evidence of Streit, mentioned above. The 1758 date comes from research developed in the 1970s and 1980s by Erich Schleier. Schleier notes[v] that the depiction of the Campo Di Rialto in one of the other two paintings in Streit’s Berlin commission shows paving that was set down only in 1758[vi].
An argument that dates two of the four paintings to 1757 would seem to require either evidence to refute Schleier’s argument for a 1758 terminus post quem (“not before” date), or to make an argument as to why the group might have been created across multiple years. All four paintings are identical in size, suggesting they were conceived as a single group, part of the same commission.
Doubts
Appealing though the notion of dating based on astronomical analysis is, I’m skeptical it can be applied successfully in this case. Arguing that San Pietro captures the moon as observed requires too much special pleading.
There are simpler practical arguments around painting processes and methods in general, and Canaletto’s in particular, that can explain what is shown.
However, we have another painting to examine — next month, we’ll take a look at La Vigilia di Santa Marta.
[i] À propos de «Jan van Eyck, les Alpes et la lune», Renzo Leonardi, Revue de L’Art n. 161/2008–3, p. 77
[ii] Venice through Canaletto’s Eyes, David Bomford and Gabriele Finaldi, National Gallery Publications, London, 1998, p. 14
[iii] Sigismund Streit, Verzeichnüß Aller Bücher, Gemählde und Andere Sachen…, Herasusgegeben und kommentiert von Peter P. Rohrlach und Susanne Knackmuß, Scherer Verlag Gmbh, 2006, p. 11
[iv] Further Adventures of the Celestial Sleuth, Donald W. Olson, Springer International Publishing AG, 2018, pp. 84–89
[v] Il Campo di Rialto, Erich Schleier, in Canaletto, Disegni — Dipinti — Incisioni, Alessandro Bettagno, Neri Pozza Editore, Vicenza 1982, p. 85
[vi] Based on commentary in the 1976 edition of Constable/Links catalogue raisonné (p. 336) which in turn references G. Tassini, Curiositá veneziane, p. 297.
