Category: Bonus!

13 October 2026

Designed by researchers at Stanford University, Palladio allows users to visualize their data through spatial maps or network analysis with just a CSV. Since we have already worked with spatial data in this course, I focused on the network analysis tool in Palladio to enormous success.

I elected to upload a pre-existing CSV dataset of Roman coins pulled from OCRE¹ to play around with Palladio. I worked with this data spatially on a project that looked at ambiguities between Roman empresses and goddesses earlier this year², but I have never considered examining the same data from a network analysis perspective. From a spatial vantage point, I wanted to see how far (geographically) coins of empresses Faustina the Elder (100-140 CE) and Faustina the Younger (130-145 CE) had spread across the Roman Empire. Running the OCRE coins datasets through QGIS revealed that the coins had spread significantly beyond the Italian peninsula (see map below).

But spatial data can only tell us so much. Running the same two datasets through Palladio’s network analysis tool allowed me to examine different aspects of the dataset in relation to one another. Particularly, I was able to plot the portrait on the obverse (the front/”head” of the coin) against the inscription on the reverse (the back/”tail” of the coin). This is significant because it allowed me to see the patterns between coins of the two Faustinas before and after they were deified³ in relation to the divine associations on the reverse of the coins. In the Roman Empire, it was common for emperors and empresses to issue coinage with their portraits (and names) on the obverse and images of gods/goddesses on the reverse–with accompanying inscriptions–to promote associations between themselves and the divine.

Using Palladio, I had the network analysis tool map the portraits on the obverse to the inscriptions on the reverse, generating the network below for Faustina the Elder. You can see the two central nodes in the network below, where coins issued before Faustina the Elder’s deification (which occurred upon her death in 140 CE) and after are shown as separate nodes. Faustina the Elder was referred to as “Faustina Augusta” initially, and later received the title “diva” (“Diva Faustina”) after her death. She died early into her tenure as Roman empress, which is indicated by the significantly larger node for “Faustina I (Diva)” in the network. I only included coins that were definitively classified as portraits of Faustina the Elder by OCRE in this analysis, so uncertain portrait coins were not included.

The connecting nodes between Faustina the Elder’s two titles indicate all of the reverse inscriptions found on coins issued both pre and post-deification. Some inscriptions are unsurprising due to their frequency across Roman coinage, such as AVGVSTA S C (Augusta senatus consultum⁴)and AETERNITAS (eternal), but others spark potential questions. Part of my original research project that started this network analysis was examining ambiguities between the two Faustinas and the goddess Juno, and strong associations can be seen in this network analysis. All instances of Faustina the Elder coinage issued with Juno inscriptions are highlighted in red in the above network, totaling six different variations across both of Faustina the Elder’s titles. Only one inscription, IVNONI REGINAE S C, is seen exclusively in the elder Faustina’s Augusta coinage. The bulk of the Juno-related inscriptions are either shared across both Faustina the Elder’s Augusta and Diva coins, or are exclusively found on her Diva coinage. This makes sense if we remember that the elder Faustina died early into her reign with Antoninius Pius, so the vast majority of coins issued with her portrait were post-deification.

Interestingly enough, the number of associations with specific Juno inscriptions is notably different between the Faustina Augusta and Diva Faustina coins. Filtering the network for only coins depicting Faustina Augusta on the obverse reveals that 114 of the 120 coins with IVNONI REGINAE were minted before Faustina the Elder’s death. Moreover, almost all of the coins with just IVNO or IVNO S C as the reverse inscription featured Diva Faustina portraits on the obverse, highlighting the shift in the elder Faustina’s association with Juno as a ruler to her association with Juno as a goddess.

Faustina the Younger is a slightly different story. The daughter of the elder Faustina, the younger empress was alive for most of her husband’s reign so the majority of coins issued with her portrait were as Faustina Augusta, not Diva Faustina. (Faustina the Younger’s husband was Marcus Aurelius, r. 161-180 CE). She had few coins issued posthumously by her husband or successive emperors. This difference from the elder Faustina can be seen in the larger node for the “Faustina II (Augusta)” title in the network analysis below.

An examination of Faustina the Younger’s network analysis also shows that the younger Faustina was associated with significantly more goddesses and titles on her reverse inscriptions than the elder Faustina was. Juno made up a much smaller overall portion of inscriptions for Faustina the Younger than Faustina the Elder, suggesting that the younger Faustina’s imperial image was not as concise or universal as her mother’s.

---

Issues & Future Implications

As amazing as Palladio seems to be to work with, it isn’t without its issues. My biggest complaint about the platform is the lack of maneuverability in the network analysis tool. Palladio generates the network from a user’s dataset, but doesn’t actually allow them to manipulate the network once it is generated. Users can zoom in and out for better visualization, but there isn’t a way to rearrange the nodes. Moreover, Palladio doesn’t currently have image download capabilities, so users have to screenshot their networks rather than download them as a JPEG, PNG, etc. Since the networks can’t be manipulated, screenshots often have to prioritize capturing the entire network over readability–as evidenced in my networks above which have some overlapping node labels.

Looking forward, I would like to generate a network for the two Faustinas and the goddesses represented on the reverse of the coins to see if certain goddesses are more common than others. This would require a more significant cleaning of my preexisting datasets, which I didn’t have the time to do for this week’s assignment. OCRE doesn’t have a field in its downloadable data dedicated to the portrait/image on the coin’s reverse, so I would have to generate that myself. There is a reverse description field, but that is far too many words and far too specific for each coin to be put in a network right now. The reverse inscriptions have been a good substitute in the meantime, since they suggest the image on the reverse of the coin.

1. OCRE: Online Coins of the Roman Empire, accessible at https://numismatics.org/ocre/. A similar database also exists for coins of the Roman Republic (Online Coins of the Roman Republic, CRRO, accessible at https://numismatics.org/crro/). ↩︎
2. See “Between Goddess and Empress: Iconographic Ambiguities in a Roman Provincial Bronze” under “Research and Projects” tab. ↩︎
3. It was common practice in the Roman Empire to deify emperors and empresses after their deaths, elevating them to the status of gods. These titles were often accompanied by official cult worship, such as the cult of Augustus or Temple of Septimius Severus at Djemlia. ↩︎
4. S.C. = senatus consultum: by decree of the [Roman] Senate. S.C. is commonly seen on Roman coins. ↩︎

2 October 2026

Bonus post!

As part of our unit on spatial data our class was tasked with creating our own maps in Google My Maps. I was originally going to use the map I started last week on Pierre-Jacques Volaire’s paintings of Vesuvius, but I realized that my project didn’t warrant spatial analysis on the scale this assignment required. I had intended to map the different vantage points from which Vesuvius was painted from in the late eighteenth and early nineteenth centuries, but I didn’t have a large enough sample dataset beyond Volaire to comprehensively achieve that. (Volaire painted from approximately four different locations from 1767-1799).

Pivoting, I instead elected to return to my roots: Pompeii.

I decided to use a dataset that already existed for this Google My Maps project, so I downloaded the data from my Roman Wall Painting Index (RWPI) collection. As of writing, there are approximately sixty paintings uploaded onto the site, representing a small fraction of Roman wall painting still in existence. My goal was to do the opposite of the RWPI: represent this corpus of wall painting spatially rather than thematically. Of course, there are already a number of sources that organize Roman wall painting by find spot including PompeiiInPictures and Pompei: Pitture e Mosaici (1990-2003), so this is by no means a new idea.

My goal was to map the paintings by wall painting style—an organizing system originally developed by archaeologist August Mau in the nineteenth century. Mau’s system sorted Roman wall painting into four categories based on their subject content, which roughly aligned with the periods in which they were painted. Mau argued for a linear development of wall painting, from the masonry style that mimicked real wall materials like marble that was popular in the 3rd-1st centuries BCE to the more intricate figural and architectural styles popular in the 1st century CE. For a more detailed timeline with examples, see here–>.

Style	Title	Dates	Notes
First	Encrustation style	150-80 BCE
Second	Architectural style	80 BCE-14 CE	End date denoted by death of Augustus (14 CE)
Third	Ornamental style	14-62 CE	End date denoted by earthquake of 62 CE
Fourth	Intricate style	62-79 CE	End date denoted by eruption of Vesuvius

The downloadable version of the RWPI data lacks spatial data, so I had to manually add that into the CSV before uploading it to Google My Maps. The find spot is listed in the data, so I went to Google Maps and pulled coordinates for the buildings that are tagged on Google (i.e. the House of the Vettii/Casa dei Vettii). For the lesser-known buildings (especially those that are not publicly accessible in the archaeological park), I had to reference published maps of Pompeii and compare it to traditional Google Maps.

This was the first major flaw of Google My Maps that I ran into. In other spatial data programs (i.e. ArcGIS Pro or QGIS), I would normally georeference a map imported into the software as a raster layer, allowing me to line up the image with its actual location on earth. Google My Maps doesn’t have the capability for georeferencing or importing raster layers, so I had to pull the coordinates the old-fashioned way.

One the paintings had spatial data, I uploaded the dataset into Google My Maps directly as a Google Sheets. You can also upload data as a CSV, but the dataset I was working with was already in Sheets format—and I don’t plan to do anything else with it—so I just kept it in G-Suite. I did have to manually split the RWPI data into two smaller datasets based on the wall painting style identified in the data, which was a little annoying. Since I intended to have two layers in my map (a “Third” and a “Fourth” style layer), I needed to upload two separate datasets into Google My Maps. For third style, see here–>. For fourth style, see here–>.

Once the paintings were placed in the map, I also went through and manually drew polygons for each of the nine regions (regio, sg.) used in the standard organization of Pompeii. Once again, Google My Maps didn’t allow for georeferencing, so I had to manually reference a map of Pompeii using one available on PompeiiInPictures. I matched up the street names within the archaeological park to the ones available on the reference map, getting the polygons as close as I could.

---

My map is visually appealing but analytically insignificant. The blue points represent the Third Style paintings and the red points are Fourth Style. From an initial glance at the map, there doesn’t seem to be a significant spatial distribution difference between the wall painting styles across the cities. The majority of paintings are concentrated in the northwestern part of the city, in regions VI and VII. There are some examples of third style farther east in regions V, IX, and I, but these are few and far between.

Much of the spatial distribution is limited by the relative incompleteness of the RWPI dataset. As of download, there are only approximately sixty paintings catalogued, which is a small subset of the total amount of wall painting that exists. The dataset is intentionally skewed towards third and fourth style painting as well, partly due to the goals of the RWPI (to catalogue figural subjects of paintings, which emerged in early third style) and the relative scarcity of earlier styles due to age. Many earlier paintings were replaced over time as tastes changed.

Many buildings in Pompeii were renovated after the earthquake that struck the region in 62 CE, leading to the seemingly large amount of late third and fourth style painting that existed at the eruption of Vesuvius in 79 CE. Moreover, the buildings in the northwestern parts of Pompeii were generally newer and owned by wealthier individuals in the city, further contributing to the larger proportion of newer painting styles present there. The western regions are also completely excavated, whereas the eastern regions of IV, V, IX, and III are still partially, if not mostly, unexcavated. New discoveries coming out of Regio IX have revealed numerous later style paintings, suggesting they are also present in large quantities there.

If the goal of my map was to see where wealth was in Pompeii, and we’re determining wealth by the presence of the most fashionable wall paintings of the time, my map would suggest that wealth is concentrated in the northwestern regions of the city. But my dataset is incomplete, and very skewed towards a certain part of the city, so maybe the presence of later styles of wall painting isn’t a great metric to understand wealth distribution in Pompeii.