Diana Kamilli: Thinking Inside the Box

Desert Archaelogy ceramicist James M. Heidke wrote this tribute to the late Diana Kamilli and her contributions to ceramic provenance studies in Arizona.

Diana Chapman Kamilli passed away in early August of last year, after a short illness. This belated Field Journal entry seeks to recognize the important contribution that she made to the methodology used in sand temper-based, binocular microscopic pottery provenance studies that have taken place in Arizona during the last 40 years.

Born in Manhattan in 1941, Diana graduated magna cum laude from Vassar College in 1963 with a Bachelor of Arts degree in geology; she then received a Master’s degree and Ph.D. in geology from Rutgers University in 1968. After a decade spent as a professor of geology at Rutgers, Vassar, City College of New York, and Wellesley College—where she served as Chair of the Geology Department for six years—she joined the Department of Anthropology at Harvard University and the History of Science Department at MIT.

Diana was hired by Desert Archaeology in 1992 to assist Elizabeth “Beth” Miksa in the petrographic analysis of sands and sherds from the Schuk Toak project area located in the Avra Valley west of Tucson. More than 100 sand samples had been collected along the length of the valley, from Three Points on the south to Red Rock on the north, but each required petrographic point-counting in order to document its composition.

Petrographic field research in the Tucson Basin. (a) Collecting Sand Sample No. 237. Beth Miksa seated on wash bank recording sample location data; (b) Sand Sample 237 coarse (left) and sand (right) fractions. Only the sand fraction was collected. The appearance of the coarse fraction was recorded in the field before leaving it behind; (c) Avra Valley petrofacies model created by Diana Kamilli for the Schuk Toak Archaeological Project pottery provenance study. Sample locations 227, 229, 230, 231, 243, and 270 lie within the project area. CLICK TO ENLARGE


Point-counting provides information on the relative volume of each rock and mineral in a sand sample. To do that, a grid is superimposed over the sample to be counted, and the composition of the grain under each point is recorded.

Diana had been trained in traditional point-counting, wherein if a point landed on a sand-sized piece of granite the sand grain would be counted as that rock type. The earliest Tucson Basin petrographic studies carried out by James “Jim” Lombard also used that method, but, under the instruction of William Dickinson at the University of Arizona Department of Geosciences, Lombard soon replaced traditional point-counting with the Gazzi-Dickinson method. Jim chose the Gazzi-Dickinson method, rather than traditional point-counting, because it alleviates statistical problems that can arise when comparing sands in various states of decomposition. According to the Gazzi-Dickinson method, grains are divided into monomineralic fragments and fine grained volcanic, sedimentary, and metamorphic lithic fragments. All monomineralic fragments are counted as the mineral phase to which they belong. All grains sand-size and larger (≥0.0625 mm) are counted.

The difference between the two point-counting methods is best understood by examining an illustration.

Schematic illustration of point-counting a granitic sand using the Gazzi-Dickinson method.


In this illustration, Points 1, 3, 11, and 12 symbolize quartz grains, Points 5, 6, and 10 plagioclase feldspar, and Point 14 a potassium feldspar. Points 10 and 11 both land on a sand-sized piece of granite composed of quartz, potassium feldspar, and plagioclase feldspar. Using traditional point-counting conventions that piece of sand would be recorded as granite. However, using the Gazzi-Dickinson method point 10 would be recorded as plagioclase feldspar and point 11 as quartz. Points 1, 3, 5, 6, 12 and 14 represent granitic constituent minerals already decomposed from the parent rock. Points 2, 4, 7, 8, 9, and 13 did not land on any sand-sized material, but, in a pottery thin section, would be counted as clay paste.

While the Gazzi-Dickinson method provides an accurate estimate of the volume of each of the minerals, it also creates a conundrum regarding the appearance of a sand in loose hand sample or in a fired clay paste: the point count data report mineral frequencies, whereas an observer looking at the sand, or sand temper, sees sand-sized granitic rock fragments as well as free minerals. To solve this problem Diana wrote detailed, hand-sample descriptions for each set of compositionally and spatially related sands, which, following geologic convention, are referred to as petrofacies. She also made representative sand grain collections for each petrofacies and mounted the sand grains in match boxes.

Three Brawley Wash grain boxes. To the left and right are Kamilli’s original boxes for her S1 (volcanic) and S2 (metamorphic) compositional subdivisions. Shown center is the current, comprehensive box. CLICK TO ENLARGE


These sand grains were illustrative of particular attributes in her petrofacies descriptions. The grains were labeled and carefully keyed to the descriptions in order to provide direct comparative reference collections for anyone using the written descriptions. Diana termed the mounted sand collections “grain boxes.”

Her innovation was quickly adopted by Desert petrographers for other Tucson Basin, Tonto Basin, and Lower Verde Valley petrofacies then under development, and in other portions of Arizona as petrofacies models were created throughout the state. A standardized format was adopted, mounting an insert with 20 numbered rectangles in each match box with representative rock and mineral grains glued in each “slot”, a summary petrofacies description written, as well as descriptions for the sand grains. For example,

  • BRAWLEY WASH PETROFACIES: Sands in this petrofacies were collected from piedmont and alluvium deposits from a highly braided and worked alluvial area. These deposits are complex geomorphically, and have been reworked and re-transported by smaller streams from the mountains. Volcanic grains (usually finer in grain size and rounded in shape) are attributed to the Roskruge and Silver Bell mountains on the west side of the wash. Granitic grains appear to be re-transported and re-deposited during events of transgression and over flooding at the northeast side of the wash.
  • Quartz (Slot 1): 20%-40%. Mostly present as lithic component, although some crystal fragments might occur individually. From clear to pale-gray, white and sallow-yellow. Luster goes from waxy to vitreous. Grains are mostly translucent to transparent, subangular to angular and they appear as subhedral and euhedral crystals.
  • Plagioclase Feldspar (Slot 2): 10%-20%. Mostly present as lithic component, isolated free-floating grains are medium sand size, often white to pale gray, subangular to angular crystals with a pearly luster.
  • Potassium Feldspar (Slot 3): 2%-10%. Mostly present as lithic component, isolated free-floating grains are medium sand size, peach to salmon in color, subangular crystals, translucent to opaque with a pearly luster.

Working as a ceramicist tasked with identifying sand temper provenance, I’ve found grain boxes to be an immensely useful tool. They facilitate rapid learning of each petrofacies’ composition and appearance. I’ve often thought that the keyed sand grain descriptions provide a metaphorical alphabet for each petrofacies, whereas their abundance in loose hand sample creates “words” from those “letters.” As well, having grain boxes and loose sands available as reference materials contributes to objective, replicable, and testable temper provenance identification.

In addition to analyzing sands and sherds for the Schuk Toak project, Diana conducted the petrographic analysis for Desert’s Country Club Road Project (Mesa), Archeological Consulting Services’ Pima Freeway Project (from the Salt River to Via Linda), Statistical Research, Inc.’s Lower Verde Archaeological Project, and SWCA Environmental Consultants’ Gibbon Springs site (northeastern Tucson). While at Desert she also analyzed pottery from the Upper Seacow Valley Area in the Karoo Basin of South Africa for Charles Bollong (Pima Community College) and assisted in the interpretation of Tanque Verde Red-on-brown pottery provenance data for Karen Harry’s Society for American Archaeology award-winning dissertation research. All ceramicists who have applied one of the nine Arizona petrofacies models while identifying sand temper provenance, or anyone else who has used the data resulting from those studies, owe a debt of gratitude to Diana for her invention of the reference grain box.

Those interested in learning more about Diana’s life and career should visit:

Diana Kamilli obituary

The Arizona Geological Society Newsletter, September 2023

A Thoroughly Modern Marriage, Science, April 1993