Rebuilding the Temple of Isis: something about doors and roofs

With the three scaled Piranesi’s drawings I was then able to calculate the dimensions of most of the main elements of the iseum.

Importing Piranesi’s maps as svg re-drawings (made in Adobe Illustrator) I could have extruded all the basic features of the iseum directly. However, I knew this process can generate a number of unwanted vertexes in 3D studio max that can be difficult and confusing to manage. For this reason, I decided to use the svg file only as a visual guideline and to redraw the plan with the line tool in 3D studio max. This choice also gave me a better control on my model.

I have simplified the shape of the columns (both the Corinthian of the main temple and the Doric of the porticus) because they were not relevant for my visualisation. If I’ll be able to take this research further, I would like to verify if the elements of the columns as Piranesi drew them, follow the rules canonised by Vitruvius.

Before building the colonnades (the temple one and the porticus one) I had to decide if I wanted to be as faithful as possible to the plan or if I’d rather to use the tools offered by the software to make a perfect alignment. Although a perfect geometric alignment is very unlikely (if not impossible) in an ancient building, I decided to apply this option to all the columns in my model because

- it was the fastest and simplest solution and the columns were not one of the focuses of my research

- I wasn’t sure if the uneven position of the columns on the plan was a reproduction of the uneven disposition of the columns in the actual temple or just material errors of the drawer.

I didn’t model any of the doors because I had not enough information about them, even though Piranesi, in his plane, designed the space for the pivots. I kept this feature for the main temple door (also because Piranesi drew interesting studies about doors and openings that I would like to include in my model in the future) but I simplified all the other doors.

I was surprised to see that according to Piranesi, the cella inside the main temple is only 1,20 meters high. I would like to verify this information on site, sooner or later.

I have also simplified the front of the temple, removing many of the architectonical decorative features. However, I believe that the peculiar shape of the Temple is still very much recognisable.

When the Temple of Isis was discovered, the roof was already destroyed. My knowledge in Roman architecture doesn’t allow me to make my own hypothesis about the appearance of the roof or the other missing elements. So I decided to follow Piranesi’s hypotheses and to apply them to my 3D model. 

In this way, Piranesi’s drawings would have been both a guideline to follow and a restoration hypothesis to challenge within a 3D environment. 

The strange case of Mr. Soane

The largest amount of information I needed for my research could be find in the third Piranesi’s drawing. It represents, on two levels, the rear elevation of the temple (including the ekklesiasterion on the back) and a lateral transection of the whole iseum.

However, when I applied the scaling operations on this drawing, I was very disappointed to see that it didn’t match the other two (the plan and the front elevation) . I double checked the process but there were no mistakes in the maths.

Although I could have just manually rescaled the mismatching drawing in order to make it fit with the rest of my model, I wanted to pursue a more rigorous approach and verify if the measurements I have derived (through geometric and algebraic operations) from the first two sources were reliable.

Consequently, I decided to cross reference the dimensions derived from Piranesi’s plan against the data recorded by John Soane.

Detail of Soane's plan fro the Temple of Isis at Pompeii.
Courtesy of the Soane Museum

As I had already noticed, the shape of the two plans (Piranesi’s and Soane’s) look quite consistent so I expected Soane’s measurements to confirm the ones I had already derived from Piranesi’s previous drawings. 

On the contrary, the numbers cited by Soane appear to be hugely different from the ones I derived from Piranesi. According to Soane, the length of the iseum's colonnade  is 46’9’’ english foot. It means circa 14,23 meters, while according to my calculation from Piranesi, the length should be 19,363 meters (x 22,795 meters).

What I was supposed to do with two sources apparently very consistent but actually so different? I wanted to identify which of the two was the most reliable. The only way to verify this information was to look for hard measurements of the iseum. Eventually, I found them cited in the book "Alla ricerca di Iside", by Stefano De Caro, a very well known Pompeian scholar. The dimensions recorded by De Caro are a very encouraging match with the ones derived by Piranesi (19,78 x 22,7 meters). Consequently, I discarded Soane’s plan and focused on Piranesi’s first two drawings. However, I am still curious to discover how is possible that such an expert and passionate architect recorded wrong measurements. 

Going back to my first problem (the little discrepancy between the first two Piranesi’s drawings and the third one), in order to complete my dissertation project I felt it was sensible to slightly rescale the third Piranesi’s drawing to make it match with the previous two that have been confirmed by modern hard measurements.

Wuthering Heights #2

In Piranesi’s plan, the scale at the bottom of the sheet is divided in 6 segments. Consequently I assumed the scale went from 0 to 6 Roman Palms.

I chose meters as my default unit in the 3DMax environment. Then I draw a box as long as the Roman Palm’s scale in Piranesi’s imported plan. It measured 3.296 meters. 

Then I built another box, measuring 6 IRF. Being an IRF = 0.2964 meters, my box measured 1.7784 mt

Finally, I calculate what percentage the scale on the Piranesi’s plan was of my known, hypothetical scale (53.95631%) and applied (through the scale tool in 3D Max) the same percentage of scaling to the entire plan.

After this operation, all the dimensions of the elements in my file where expressed in meters and easily measurable.

I repeated the same operation with the frontal elevation. The scale Piranesi used this time was 20 roman palm. So I calculate what is equivalent in meters of 20 IRF and tried to rescale the drawings. Theoretically, the two drawings, once rescaled according to IRF and translated in meters, were supposed to have consistent dimensions. I was surprised to  see that the front elevation was much bigger than the plan and couldn’t possibly fit.

The reason was very simple. The 6 segments on the main plan stood for 60 and not 6 palms. I rescaled the first plan in 3D Max ten times its dimension (1000%). After this adjustment, the two drawings fitted quite well. This consistency allowed me to use the front elevation to derive the heights of some elements such as the core components of the temple and the four altars (two for side). 

Wuthering Heights #1

As I have recently found out, the troubles in 3D visualisation of ancient buildings mostly come with the heights. It is relatively easy to find architectural plans, especially of very well known historical places such as Pompeii or Herculaneum. But what about heights? How high were those columns that now appear like little circles on a 2D drawing? Apparently, recording the heights was not a primary concern of those 18th century archaeologists and architects who had the privilege to see the Temple when it was still very well preserved. 

Luckily, I have found the very precious Soane’s elevation and, even better, the rich Piranesi’s visualisations that I have chosen as my main reference to derive the heights.

Rear elevation and transection of the Temple of Isis in Pompei
by F. Piranesi. Accessible at Bildindex

I assumed that three drawings made by the same author and based on the same unit should have showed an high level of consistency. I was definitely over optimistic...

Piranesi included two scales in each of his Pompeian architectural drawings: Roman Palms and Piedes de Paris. For the purpose of my research, it was not really relevant which one I chose, as long as I was consistent with my choice. However, being Piedes the Paris a very tricky and slightly uncertain unit, I preferred to use Roman Palms.

The three drawings (the plan, the front elevation and the drawing showing on different levels the rear elevation and the lateral section) share the same unit system but the author applied different scales according to the dimensions of the sheet and the depicted subject.

In order to unify the dimensions and bring them all into a metric system, I had to “translate” them into an external known measure and rescale them proportionally. After that, I will only had to measure the heights of the elements in Piranesi’s drawings and build them in 3D Max (or so I thought...)

Even though the Pompeian Temple of Isis was probably built in Oscan foot, I decided to use he more established IRF (Imperial Roman Foot).

I started with the main plan. I have opened the jpeg file downloaded from the Bildindex in Adobe Illustrator and I have saved it in svg (the vectorial format). Then, using the pen tool I have re-drawn all the lines. Finally, I have exported the file in dwg (the autocad format) in order to import it in 3D Max.

The file was smoothly imported by 3D Max and the lines were potentially ready to be extruded. However, I prefer to use the imported lines only as a guide and keep an higher control on the model.

Even though Piranesi was an experienced architect, the digital plan of the Temple  I produced should be considered only as a reliable approximation due to the many variables that can have affected its accuracy.

In the first place there are all the mentioned factors that can easily interfere with the material process of drawing (and that make almost impossible to find two identical records of the same place or building). Then, both the processes of engraving and printing could have added a level of information corruption.

Form the digital point of view, all the operations of scanning,  compressing in jpg, uploading on the server, downloading from the server, converting in svg, exporting in dwg, importing in 3D max can potentially add a new level of data corruption.

Moreover, the biggest issue in the reliability of this information is probably my re-drawing of the plan in Illustrator. As almost all the ancient buildings (and many of the modern ones), the Temple of Isis is not made of perfect, straight lines. Developing only a simplified model as proof of concept for my dissertation, I decided to slightly modify the lines in order to obtain a more geometrically regular plan.

Ancient Transparencies

Plan of the Temple of Isis by F. Piranesi
Accessible at bildindex

In the end, I have decided to use Piranesi’s plan as main reference mostly because of:

- the presence of the ekklesisterium

- the availability of Piranesi’ detailed drawings of the front and rear elevation of the temple and a longitudinal section of the entire Iseum (all expressed in the same unit of the plan).

However, I intend to use Soane’s plan to double-check the distance between the elements and, possibly, to derive the height of the purgatorium. Curiously, the latter has been ignored by the meticulous Piranesi and does not appear neither in the section or the elevations he drew (or, at least, it does not appear in the drawings I have found so far).

One additional reason is that I was really impressed by the visual code Piranesi adopted in his drawings. If Soane drew and recorded the appearance of the Temple as it was when he saw it, Piranesi tried to imagine how the temple might have looked like before ist destruction. In other words, Piranesi developed an interesting attempt of 2D visualisation showing fascinating similarities with modern applied visualisation research.

Of course, Francesco Piranesi was not the first person to draw an hypothetical restored version of the Pompeian ruins. But the stress should go on the word “hypothetical”. Many of his contemporaries (and many of his modern times successors as well) simply disseminated reconstructions of the damaged building, without giving the audience any means to distinguish between original remains, grounded hypotheses and artistic imagination. As many past and present reconstructions, they were misleadingly presented as “the image of the real thing”.

On the contrary, Piranesi shows a praiseworthy transparency making immediately recognisable the different status of the drawn elements.

Front elevation of the Temple of Isis by F. Piranesi
Accessible at bildindex

In my opinion, his code is simple and effective even if it is expressed in grey scale.

All the elements actually still present in situ when the Italian architect and engraver drew them are represented in a realistic and detailed style.

The missing elements or, in other words, the one virtually restored by Piranesi are represented in an essential, graphic style: few main lines, no shadows, no chiaroscuro.

Using this opposition, Piranesi produces a document that, at the same time, gives an idea of the original look of the building and witnesses its actual state, giving to an expert audience the opportunity to agree or disagree with the artist hypotheses.

The striking use of white for the restored elements underlines the concept of absence and their status of “ideas” and not “facts” within the visualisation.

My knowledge of Roman first century AD architecture does not allow me to evaluate the likelihood of Piranesi’s restoration. He does not mark the level of certainty of his hypotheses and do not motivate them. However, his concern about transparency and his scholarly attitude can be considered an early example of best practice and his basic code seems to me very powerful from a communicative point of view and definitely still valid.

The thickness of things

Before deciding which plan I will have used as main starting point for my modelling, I wanted to check their consistency. I created a file with Adobe Photoshop and, after reducing the opacity of the three plans (Saint-Non's, Piranesi's, Soane's) I put each one on top of the other.

I have chosen one single element to compare the three plans (the right side of the colonnade, watching the Temple from the front) and rescaled them in Photoshop until that element was the same in all the three layers. 

LEFT: Piranesi and Soane's plans RIGHT: Saint-Non and Soane's plans

This superimposition has showed that none of the maps is a perfect match with any other. 

However, Piranesi’s map and Soane’s one have a certain degree of consistency while Saint-non’s map appears to be remarkably different from both of the other two.

I believe this discrepancy does not disqualify any of sources. These kind of differences can be due to many variables including the thickness of the pencil, the texture of the paper, human error in measuring, transcribing or drawings or simply the very difficult task of taking measures of such an irregular and massive building as a partly damaged Roman temple.
Moreover, working not with the originals but with digital copies, we have to take into account an other potential level of data corruption.

Plans for the Past

Plan of the Temple of Isis by J.C. Richard

The next step in order to build my 3D model of the Pompeian Iseum is to gather reliable measurements. The ideal condition would probably be going in situ and record the actual dimensions of the elements.

Even though I took extensive photographic documentation, when I went to Pompeii I had neither the equipment or the expertise necessary to undertake personally the operations. So, I looked for early records of the excavation and architectonical plans of the temple’s area. The old royal prohibition of documenting the site and the finds, makes not particularly easy to find detailed documentation. However, some of the authorised records have been very well preserved.

The first map I chose is the one drawn by Jean-Claude Richard, (aka the abbe de Saint-Non) in 1782 (Voyage Pittoresque ou Description des Royaumes De Naples Et De Sicile). The plan is very neat and includes the area of the ekklesiasterium in which I am particularly interested.

However, Saint-Non’s plan included no information about the height of the architectonical elements. Without at least one known measurement, it was impossible for me to start building a reliable model. So I kept looking for more resources. 

I was lucky enough to stumble on the digital archive of the Sir John Soane's Museum in London. Two of the documents in the Soane’s drawings collection were extremely interesting to my research: a plan of the Temple of Isis and a close up of the purgatorium. Promisingly, Soane included in his drawing a frontal elevation of the temple, as it looked like in 1779.

Plan of the Temple of Isis by J. Soane

I went to the museum to examine the drawings. It was two times exciting. First because I had the opportunity to handle beautiful and accurate documentation more than two hundreds year old. Secondly because the Soane Museum is itself a unique and wonderful place.

This is probably the right moment to thank again the Museum because when my supervisor and I explained them the topic of my research they kindly offered me an high quality digital copy of the drawings.

With the digital files I was allowed to make some experiments. First of all checking (using Adobe Illustrator) if the elevation drawn in the upper part of the plan shared the same scale of the plan. If so, I could have use it to derive the heights of some of the main elements of the temple. I have chosen two elements (the diameter of the columns and the length of the temple frontal footstep) and I have measured them in both the plan and the sketch of the elevation. As they matched, I had eventually a starting point for my model. Moreover, the source was one of the most famous and gifted architect of this country.

An additional bonus in Soane’s plan was the presence of a very accurate documentation of the distances between architectonical elements. The only drawback was the absence of the ekklesiasterium area, which is crucial to my research.

Few days ago, I was suggested to visit a very rich German digital resource on ancient architecture: Bildindex der Kunst und Architektur. I used “Isis” as key for my query and I found several relevant documents, mostly created by architect and engraver Francesco Piranesi.

Not only Piranesi produced an other plan of the Iseum, but he also drew very detailed elevations of the entire complex. Both plan and elevations were expressed in the same double unit (Roman Palms and Piedes de Paris), so it would have been relatively easy to derive the heights from there.

Eventually, I had three different plans: each of them interesting and rich in information, each of them created by a re-known and reliable source.