An enigmatic circular engraving of the view from St Pauls’ Golden Gallery


[Note: the illustrations to be added]

In The Map Collector for June 1987 Ralph Hyde of the City of London Guildhall Library described an anonymous and undated circular view of the City - the Fisheye View.  This curiosity compacts the full view in all directions from the top of St Paul's Cathedral into a 74 centimetre diameter circle, with a central 30 centimetre diameter blank portion containing a key and a perspective of the Cathedral.

Only three examples are known, but Ralph Hyde convincingly argued that the Fisheye View was derived about 1845 from the lost cylindrical painting by Thomas Horner, exhibited in the Colosseum in Regents Park in 1829-1863. Horner had spent three  years from 1820-3, laterly in a rotating shed (possibly a camera obscura from the curious angle of the roof) on scaffolding erected above the golden cross of St Pauls preparing the 300 working drawings.

The article went on to point out that the Fisheye View was in  fact  a particular type of a  previously  known construction, the anamorph.  This is a perspective which uses a picture plane not at right angles to the direction of view, and which may be curved. The result is that a correct reconstruction can only be obtained from a single viewing position, sometimes reflecting the image in a mirror, flat or curved; from any other position the picture appears grossly distorted and in most cases unrecognisable.  They enjoyed a minor popularity as puzzle pictures or for pornography and treasonable portraits. The National Gallery has just cleaned and re-exhibited an excellent Perspective Cabinet by Samuel van Hoogstraten (c1657) which demonstrates the effect admirably, and may be available as a fold up cardboard model in the near future.

The one question left open was what the Fisheye View was for. It is rather large and cumbersome, difficult to look at as it requires turning constantly and would be ruined by being folded - it was suggested that it might have been designed to fit into some kind of viewing cabinet. In this article I shall present some possibilities.

The first thing to strike one is the size of the central blank area, much too large for a conventional cylindrical mirror anamorph - if cut out the hole is perfectly sized for putting one's head through, thus wearing the View as a kind of horizontal Elizabethan ruff.

Looking down, assuming one's close-up vision is adequate, one gets a surprisingly realistic effect, and can rotate the View to see behind. Figure 2 is an analysis of the fisheye transformation used and reveals that the effect is to apparently bend the real ground surface downwards towards the horizon into a round-topped conical dome, with the observer looking vertically downward at the level top, and the horizon round the circumference of the base. All verticals in the original scene transform to radial lines pointing to the centre of the circle, and rectangles (for example buildings) therefore come out wedge shaped - the nearer they are the worse the distortion.  Modern fisheye 180 degree camera lenses produce exacly the same effect.

What  is  unexpected is that this distortion is  not significant over a small area when viewed from the centre of the View in the manner suggested. This is because if, in real life, one looks downwards all verticals converge to a point directly below one's feet - just as if one looks up they converge to a point directly above one's head (the building-leaning-over-backward  effect  of unskilled photographers).  An  inherent feature of the fisheye transformation is that all verticals become radial lines pointing directly at the centre and in the suggested viewing position this is optically identical to a point directly below the feet; they are therefore read as true verticals.

Figure 1 demonstrates this effect on a simple cube. Viewed normally it appears wedge shaped and elongated, with the verticals converging to a point just above the bottom edge of the paper - exactly like a building near the centre of the Fisheye View. This is because it is an anamorph, using a horizontal picture plane (the page), and a  close view-point (marked 'cube view position'). If you hold the page up horizontally and ledge it between your chin and mouth, with one eye directly over the viewing position (close the other eye), you will see a perfect cube sitting on the page, its edges apparently vertical.

That  this works for the Fisheye View is  completely fortuitous, but may have been the only method of viewing envisaged; it is certainly very easy, if looking a little ridiculous - unfortunatly one still gets the impression of looking down onto a conical dome, with the horizon well below the horizontal.

More interesting is to devise a method of reconstrucing a true 360 degree panorama all round one, with the horizon truly horizontal, as suggested by Ralph Hyde. The View is not laterally inverted, so this would involve two mirrors, and a possible configuration is shown in section in Figure 3.

The horizontal line of view is turned through two 45 degree angles to drop vertically onto the horizon of the View. As the View is circular these two mirrors (angled much like a pentaprism in a single lens reflex camera) have to be curved round it to form truncated cones. The observer inserts his head from below through the central hole as before.  Some method of lighting the View is neccessary, and a window in the angle between the two mirrors is suggested. This also has to be curved into a truncated cone.

The finished object is rather complex and difficult to constuct,  nothing being rectangular, and  the  exact dimentions were arrived at by graphical trial and error. Accurate conical mirrors in particular are a substantial technical problem. It is thus open to question whether such a device could have been constucted at all by the producers of the Fisheye View, and it is certain it could not have been economically mass produced, as the mirrors would have had to be made from highly polished speculum metal. One could perhaps envisage a single example as an attraction in an optics gallery, much like Hoogstraten's Cabinet.

A rough half scale version has been constucted out of cardboard and mirrored Melinex plastic film to check that the effect is correct. The result, though not very good due to problems with the mirror fixings distorting the mirror, were sufficient to prove the principal. One problem could not be assessed, and that is the effect of binocular vision. Ideally the viewing position should be a single point (as in figure 1), and it is difficult to know if having two eyes either side the ideal point would produce a confusing effect in the full size version. The further back the viewing position is the less this is important, but the current design in fact puts the observer as far forward as possible in order to get the closest (and therefore largest) image and so that when the head is turned the axis of rotation is at the centre of the circle.

A full sized professionally constructed version would give a unique time-slip into London's Victorian past from an unatainable viewing position (one can no longer peer out from beneath the Golden Ball since the recent renovation). A comparison fisheye photograph could also be attempted of a view which has transformed since the blitz.

However, without some forgotten attic disgorging a tarnished set of truncated cones, along perhaps with Horner's original working drawings, it is impossible to say if the Fisheye Viewer was ever originally constucted - only that it could have been.

Copyright 1988. Roger Morgan is an Architect with an interest in anamorphs and perspective.