(1) Overview

Context and motivation

The Gandhara palettes (also known as toilet-trays: ) are a distinct category of artefacts hailing from the region of Gandhara in North-western Pakistan (Figure 1). Morphologically, the palettes are small concave dishes, usually averaging around 4x19 cm in size (; ; Figure 1). Only two depositories of known samples have been compiled (; ), making a comprehensive study of their iconographic patterns and relation to Gandharan art difficult. This paper, based on my MA Dissertation at UCL, describes an updated database of the palettes and provides statistical data which will help future studies of iconography and hybridity in the ancient world. I provide information on the palettes’ iconography and materials, and bibliography for future discussion on their artistic contributions (; ).

Figure 1 

Indo-Greek banquet with Satyrs (Ancient Orient Museum, ID 81 in Database, https://en.wikipedia.org/wiki/File:IndoGreekFestivities.JPG) It’s from Wikipedia CC-BY 3.0.

(2) Method

I have built a database based on museum entries from museums such as the Missouri Museum (; ), the British Museum (), and Taxila Museum (; ). The first stage includes collecting all the data into one table. Variables include Diameter, Provenance, Description, Period, Phasing, to put the material into statistically relevant categories, as well as various forms of iconographic descriptions based on different mythological or functional descriptions. I then use the database to structure queries about the palettes’ spatiotemporal and functional distribution. Other information such as chronology and material can also provide more insight into qualitative and quantitative studies on the dataset in the future.


I am building upon Dar () as well as Francfort () and their methodologies by building an analogous list. These methodology lists were compiled before Erdosy’s () updates of Taxila’s stratigraphy as well as Lo Muzio’s updates on the palettes’ chronology (). Therefore, previous lists need chronological re-appraisal. I have opted to adhere to the museum chronologies, many of which span two centuries. The large quantity of unprovenanced finds further complicates chronology (). As such, I rely on their museum recorded chronologies, as these are the best possible selections. The use of a relational database is my attempt at innovation by bringing up new categorizations to understand their distribution and their different artistic contributions.

I have used online museum catalogues such as the Metropolitan Museum of Art, the Ashmolean Museum, and the British Museum as well as their publications (; ), and built a database, organising entries by discernible iconographic themes within the decorative detail. I have organised them chronologically and made use of the location and provenance to see if they were found in religious or domestic contexts (Table 1 and Table 2). To avoid forgeries, I have used museum records to check the themes that have been recorded in archaeological contexts to further ensure accuracy. Falk’s () analysis of chronology is also helpful, as now we can date the earliest trays in Taxila at around 80 BCE until the 3rd Century CE. I have left fragmentary finds in my database to build a more complete picture of the palettes’ spatiotemporal distribution (see Tables 1 and 2). The relational database is my major innovation in terms of analysing the palettes.

Table 1

Provenance distribution of the palettes.


Narai2c. 1%

Rawalpindi1c. 0.5%

Akra2c. 1%

Barikot1c. 0.5%

Sahri Bahlol1c. 0.5%

Sar Dheri1c. 0.5%

Charsadda2c. 1%

Utmanzai1c. 0.5%

Taxila (unclear mounds)56c. 29%

Sirkap7c. 3.6%

Hatra1c. 0.5%

Jota Khan1c. 0.5%

Udegram3c. 1.5%

Butkara1c. 0.5%

Barama1c. 0.5%

Tor Dher1c. 0.5%

Malka Dheri1c. 0.5%

Uddiyana3c. 1.5%

Unknown112c. 43%

Total196c. 100%

Table 2

Chronological distribution of the palettes and phasing.


2nd– 1stBCE 22c. 11.4%Phase 1

1stBCE – 1stCE 39c. 20.1%Phase 2

1st– 2ndCE 53c. 27.3%Phase 3

2nd– 3rdCE 9c. 4.7%Phase 4

3rd– 5thCE 5c. 2.6%Phase 5

5th– 7thCE 4c. 2.1%Phase 6

Un105c. 53.3%Unknown

Total196c. 100%N/A

Sampling strategy

To compensate for these issues, I have taken a comprehensive approach to analysing the iconography. The database has multiple columns with information from museums and scholarship, which is combined with my categories to describe the palettes. One column pertains to Museum/Scholarship descriptions either in purely narrative or in mythological terms. These vary extensively, because they include both functional and mythological themes, as well as because different scholars make them. There is a Fragmentary column to acknowledge which entries are Complete, Partially or Fully Fragmentary. There is also a column describing what type of compartments exist, if at all, in a palette. The Other column includes other information such as backside design, floral patterns, or carving techniques; this includes museum information where that is available, which can be a useful cross-reference for which figures appear in palettes. I also opted to use Lo Muzio () and place the 1st century CE as a baseline for the palettes’ presence in Gandhara. In the case of the palettes, I have used the museum chronologies, which often have a range of two centuries, and combined them with imagery and stratigraphy to build a temporal distribution pattern for the palettes (Table 2).

Quality control

To make categorizations easier, I have created three General Description columns and three Mythological Description Columns. The General Description columns help build meaningful categories to study the palettes. The Figures column describes the figures that are involved, whether they are humans or animals. The Activity column describes what these figures are doing, and the Context Column describes the background based on furniture or the presence/absence of flora and fauna. The descriptions are my own and describe the activities involved. To build the queries, I have also added two more columns that complement the Environment Column. One focuses on the narrative that is produced (i.e., whether the scene shows chasing or fighting) and the other focuses on behaviours in the narrative (e.g., whether a scene is erotic or has wine, as well as what the figure is if no activity is taking place). The Museum Chronology category includes chronologies recorded in museum entries. I have also created a Phasing column to put the chronologies into organized phases. The description of the palettes is important because of the presence of foreign elements for elite production (), which is the reason why the Mythological columns were created. I have collected various interpretations for motifs from Greek, Persian, or Indian religious traditions. Many interpretations are taken from scholarship, but I have also used my own based on analogies. I have also provided citations of the texts they were found in, such as museum catalogues, in the database in a separate column.

(3) Dataset description

Object name – PalettesDatabase

Format names and versions – .accdb, .xslx, .csv

Creation dates – 05/01/2020 – 05/10/2020

Dataset creatorsChristos Nikolaou


License – CC – BY 4.0


Repository name – Apollo, University of Cambridge

Publication date – 20/03/2023

(4) Reuse Potential

The total of 195 entries has ample information on variables like material, compartments, imagery, provenance, and chronology. The incomplete nature of many records makes analysis quite difficult. However, there is ample information on material, and a majority of palettes are provenanced and have their cavities described (See Tables 3 and 4).

Table 3

Material distribution of the palettes.


Ivory1c. 0.5%

Silver2c. 1%

Copper1c. 0.5%

Bronze1c. 0.5%

Schist38c. 19.7%

Serpentine2c. 1%

Steatite16c. 8.3%

Stone18c. 9.3%

Terracotta3c. 1.5%

Unknown112c. 57%

Total196c. 100%

Table 4

Compartment distribution of the palettes.


Crescent Shape2713.8%

Cross Shape94.6%

Image Shape3015.4%



Rock Shape21%

Sun shape10.5%

Swastika Shape10.5%

Bowl Shape31.5%




There are also future possibilities in terms of more qualitative analyses. For instance, I plan in future articles to examine whether specific compartments fit to specific artistic tropes, or the regional and spatial spread of tropes like Horse Riders, and then compare different Horse Rider examples between them or with other similar scenes like Sea Monsters. For instance, by seeing how iconography may change based on different material, or whether specific tropes appear in specific periods, we can better understand the artistic and social shifts that happened in Gandhara through the art which was commissioned. Relational databases are also useful for putting artefacts into neat statistical categories for combined qualitative and quantitative research and can help with mapping if coordinates are utilized.