Cast shadows are often the last elements to be added to a painting, and are readily forgotten when a plein air painting has overrun. Although that is seldom a disaster in itself, it can result in shadows being hurriedly painted in at the very end of the outdoor session, when they may be quite different from others which were painted in earlier.

Fortunately omission of shadows is less often noticed than most painters fear. However marked inconsistencies in shadows can be very easy to notice, and should be avoided at all costs. I here look at three cases in which cast shadows have gone awry or absent, before moving on to one of the great controversies in painting: whether cast shadows have colour.

The great Canaletto is known for his paintings of the canals and buildings of Venice. Careful study of his works shows that he didn’t paint what he would actually have seen, and frequently made ‘adjustments’ in his views to satisfy his aesthetic expectations. What’s more, his depiction of shadows gives insight into the compromises which he made when painting a view.

A good example is his The Entrance to the Grand Canal and the Church of Santa Maria della Salute, Venice, made between 1735-40. He shows the shadows cast by the church at the right and the figures around it: the light source is the sun, which is at the right and at an elevation of around 45 degrees above the horizon. Cast shadows therefore fall from right to left, in the picture plane.

Look, though, at any of the many gondolas on the water, such as that near the centre, closest to the viewer. Its reflection is shown clearly in the water – including those of the two occupants with their golden yellow clothing – but neither the boat nor its occupants show any shadow cast to its left.

Canaletto knew the buildings well, and would have compiled preparatory sketches and studies, in which shadows would have been included, or would be easy to add according to the time of day. Adding them for figures is also quite straightforward. In general, the gondolas shown are staffage, invented and added purely in the studio, where their shadows would be more difficult to estimate. In addition, showing them with both reflections and shadows was significantly more difficult, and potentially confusing. Yet few viewers ever noticed their absence.

For Paul Cézanne learning to paint en plein air alongside Pissarro’s easel in 1873, cast shadows were not left until last, but painted as he completed each section of this view of the House of Père Lacroix, Auvers-sur-Oise. However, as with all beginners, he took a long time getting the painting to look right, so different sections of the roof were painted several hours apart.

As shown in the marked-up image below, the angle subtended by the shadows is not consistent. In some, the sun is high in the sky, close to its zenith, but in others rather lower. These imply a wide range of solar elevations, an issue which often affects those still learning how to paint plein air.

Early plein air painters in the Roman Campagna knew that they had to capture their motifs as quickly as possible, as the light changed. If you take more than an hour or so to paint a view, you have to make a decision as to when you depict light conditions and shadows, and ‘freeze’ the image in your mind at that point. If you do not, as time passes, so the light and shadows change, and your painting becomes internally inconsistent.

As his practical skills improved, Cézanne, like Sisley and some others, tended to avoid painting most cast shadows altogether.

Another famous artist who had a struggle with cast shadows is Georges Seurat, in his famous Sunday Afternoon on the Island of La Grande Jatte from 1884-6, whose painstaking deliberation is the antithesis of the plein air view. During the hundreds of hours which Seurat spent applying small marks of paint to construct this monumental work of Divisionism, his lighting conditions wandered.

The woman standing fishing at the far left of the painting casts her shadow at an approximate angle (viewed from above) of 0130-0200 by the clock face, and hers is quite a long shadow too. Follow her shadow back until you meet a man standing wearing a ‘pith’ helmet, and he casts almost no shadow, as if the sun was almost vertically above him.

Continue to the right and towards the foreground and there is a woman holding a red parasol, walking towards the viewer, with a small girl beside her. Their shadows are cast in a different direction, closer to 0230-0300. Although most of the figures beyond the immediate, shaded foreground are seen to cast their own shadows, there are almost no corresponding shadows cast by the trees closer to the water, leaving most of the grassy area in unbroken sunlight.

Seurat also entered the fray in one of the great artistic debates of the late nineteenth century: what colour are shadows? Throughout this painting, he shows cast shadows with reduced lightness and chromatic shift towards blue, resulting in a deeper, bluer green.

Look carefully in the paintings of the old Masters, such as Poussin, and you will see that their shadows rarely show any significant chroma, and reflected or cast lighting effects are very seldom shown.

Although most of the still life paintings by specialist Chardin follow that tradition, some, such as his Copper Water Tank, have visible colouration in their shadows. In this case, it could be argued that the reflective surface of the copper tank would be responsible.

From the early days of Impressionism, several of those now considered to be Impressionists added colour to their shadows. Pissarro did so in this marvellous painting of Hoar frost at Ennery in 1873. The rhythmic cast shadows of trees are here dark brown where they fall on the ploughed area, and blue-green further back where they fall on frost-covered grass.

In Pissarro’s later Saint-Charles, Éragny of 1891, areas of grass which lie in the shadow of the trees are intensely green, whilst those in full light are gold.

In Cézanne’s Still Life, shadowed areas of the tabletop, white bowl, and other parts have quite different colours. This is even more obvious in the detail shown of Signac’s Demolisher below.

This change in approach is not confined to those who might now be considered to be core members of the Impressionists: John Singer Sargent’s oil sketch of An Artist in His Studio from 1904 depicts the shadows in its crumpled sheets as being pale brown and blue, at the right.

Frederick Carl Frieseke, who excelled at paintings of dappled sunlight such as this Nude in Dappled Sunlight from 1915, includes a gamut of different colours in the shadows cast on this nude, including browns, pink, golden yellow, blue-green, and purple.

It was the Canadian landscape painter Tom Thomson who perhaps showed the most richly-coloured shadows, cast here on snow: seen in his Snow in October (1916-17), above, and Early Spring, Canoe Lake (1917), below.

It would be absurd to suggest that these differences result from changes that have occurred in the physical world, or that these artists had quite different visual systems, or powers of observation. But there is one well-known psychophysical effect that must be borne in mind, that of colour constancy.

If you view an object with a distinctive known colour, such as an apple or orange, under a wide range of different light, with different hues and intensities, the object tends to be perceived in most conditions as having the same colour as in white light. This perceptual constancy is lost in more extreme lighting conditions, though.

If you were to measure the colour of the object objectively, then it would of course vary much more than our perception. But the painter is not just concerned with mimesis. It is not hard to make an orange really appear brown, or almost black. Should the artist then paint it according to its ‘true’ colour, what we would perceive given colour constancy, or the colour which would help us recognise that object most easily?

That, surely, is an artistic choice, not a matter of right or wrong.