Chemical Analysis of the Contents of Ancient Kohl Pots from the Nubian Kingdom

Chemical Analysis of the Contents of Ancient Kohl Pots from the Nubian Kingdom Einar Lidén

Degree project C in chemistry Department of Chemistry, BMC, University of Uppsala

Supervisors: Prof. Jonas Bergquist,

Emma Hocker, Senior Conservator at Gustavianum

1

Chemical Analysis of the Content in Ancient Kohl Pots from the Nubian

Kingdom

Abstract

Beauty is subjective and since the dawn of time mankind has tried to enhance its beauty in various ways. In our modern culture one very common practice is makeup, but some kinds of makeup have been around for thousands of years – especially kohl. Kohl is a makeup that was used in ancient Egypt and Nubia and it is still being used today, possibly thanks to Prophet Mohammed’s blessing. But what is it? What does it contain? This study aims to answer these questions, using ancient samples from the Nubian kingdom. This with the use of XRD, SEM- EDS and ICP-AES. The ancient makeup consisted of lead (up to 90%) and sulphur (up to 17%), along with 11 other elements. As far as structure goes PbS, Pb 2 O(SO 4 ) and SiO 2

crystals were identified.

2

Table of Contents

Abstract ... 1

Abbreviations ... 3

Background ... 4

Experimental ... 9

Results ... 13

Discussion ... 21

Thanks to ... 24

References ... 25

Appendix I Samples & Sample Preparation ... 27

Appendix II – Sem & SEM-EDS data ... 28

3

Abbreviations

Abbreviation Meaning

BSE Background Scatter Electron

FAAS Flame Atom Absorption Spectroscopy

ICP-AES Inductively Coupled Plasma - Atomic Emission Spectroscopy ICP-MS Inductively Coupled Plasma - Mass Spectroscopy

MVA Multivariate analysis

ND Not Detected

PCA Principal Component Analysis Raman Raman spectroscopy

SE Secondary Electron

SEM-EDS Scanning Electron Microscopy - Energy Dispersive Spectroscopy SJN The Scandinavian Joint Expedition to Sudanese Nubia project UNESCO United Nations Educational, Scientific and Cultural Organization XRD X-Ray Diffraction

XRF X-Ray fluorescence

4

Background

Beauty is in the eye of the beholder. What is considered beautiful changes over time and space, with seemingly one constant – enhancements of human characteristics that are considered beautiful.

In our modern society this can be seen everywhere, from cloth fashion to makeup and even body modifications. What is interesting is that all types of beauty enhancements can be seen in very old cultures. For example, in ancient China, small girls` feet could be bound to constrict growth, giving a more desirable look, a practice that lasted into the early 20th century (Ko, 2005). In various tribes rings could be placed on people’s necks to stretch them out (Waddington, 2002). There is also evidence that makeup has been used for thousands of years (Scott, 2016). For the earliest makeup it is difficult to tell if it was used for beauty enhancement or spiritual purpose. Given that evidence for this type of human behaviour can be found all throughout our common history and, importantly, on different continents we will probably continue to enhance our appearance even in the future.

This raises a question: why have people tried to enhance their beauty for thousands of years? I postulate that this is done to boost the chance of attracting partners and thus increasing the chance of gaining more and healthier children. I would also think the more extreme body modifications could be compared to sexual selection (Crawford & Krebs, 2013) that has run amok. Take a culture in which large lip piercings are common. It most likely started out with a small piercing, and then it was noticed that those who had such piercings attracted more partners. Then someone figured that having a larger one would attract even more sexual partners and so in small increments a modest piercing developed into a huge piece of wood imbedded in the lip.

As far as makeup goes, it has been used differently, from subtle touches to body painting to warpaint.

Kohl

This paper will focus on one type of eye makeup, kohl (see figure 1 for a typical kohl sample as well as a possibly contaminated sample). Kohl is an ancient form of makeup that was widely used in ancient Egypt, Nubia and the surrounding areas (Genge, 1979). It is a mineral based product, that very often has a base of galena and thus a lead base. Due to the

construction of the High Dam in Egypt a vast land area was submerged. To preserve the

cultural heritage of the region a huge UNESCO (United Nations Educational, Scientific and

Cultural Organization) project was launched to salvage as much as possible. As a part of this

project The Scandinavian Joint Expedition to Sudanese Nubia project (SJN) was formed by

the Nordic countries that excavated a 375 km stretch along the east bank of the River Nile at

the Egypt/Sudan border. This was the origins of the samples that were used in this project

which have since been stored at Gustavianum in Uppsala (Genge, 1979).

5

Figure 1 A typical sample of kohl powder to the left and a sample of possibly contaminated kohl sample to the right.

The chemical makeup of kohl differs from place to place. Some types seem to be pure galena (PbS) whereas other kohl is made with wet chemistry. Tapsoba et al. (2010) analysed, using XRD (X-ray Difraction) and electron microscope, over 50 kohl samples from the Louvre and found that they contained galena (PbS), cerussite (PbCO 3 ); phosgenite (Pb 2 Cl 2 CO 3 ) and laurionite (Pb(OH)Cl). Since phosgenite is not naturally occurring in the region, the authors described a way to synthesise it, see figure 2 (Tapsoba et al., 2010).

Figure 2. A way to synthesise phosgenite, possibly used in ancient Egypt (Tapsoba et al., 2010).

Even though kohl has an ancient history, it is still being used today. Unfortunately, nearly half (48% (Hardy et al., 2002)) of modern day kohl has a base of galena, lead (II) sulphide,

meaning it still contains high levels of lead. Analyses of modern khol whit Jarrell-Ash Plasma fission spectrography has shown as much as 69% lead (Parry & Eaton, 1991). The continuous use of traditional kohl in the Middle East could be explained by both tradition and religion. In Sunnan Abu Dawud, which is an account of the prophet Mohammed’s life where his values are described (or Ḥadīth), from which religious laws and moral are derived, Mohammed advocates the use of kohl (lmâm Hâfiz Abu Dawud Sulaiman bin Ash'ath et al., 2008). Since Sunnan Abu Dawud is a supplementary text to the Koran, conservative Muslims continue to use kohl, even though it contains high amounts of lead. This shows how deeply kohl is embedded in the Middle Eastern culture.

Analytical methods.

The literature of analytical chemistry on kohl composition is limited. There are however some

methods described in the literature. For example, 18 modern samples from the Abu Dhabi

market were analysed with SEM (Scanning Electron Microscope), SEM-EDS (SEM-Energy

Dispersive Spectroscopy) as well as XRD. Eleven of these samples were lead based and

contained a major phase of galena (up to 90%). The minor phases were cerussite, anglesite

and one sample contained zincite as a minor phase. Of the remaining nine samples five had

zincite as a major phase. Three samples contained either Sassolite or amorphous carbon. The

major phase of the final samples was calcium carbonate. SEM-EDS determined that the

samples mainly contained: Pb, S, Mg, Ca, K, O, Zn, Si, Na, Al, where Pb and S were mainly

found in lead based samples and Zn was almost exclusively found in samples with zincite as a

6 major phase (Hardy et al., 2002). Four years later, another study was made, which also used XRD and SEM on 18 new and six ancient Egyptian kohl pots. XRD revealed that six of the modern-day samples had a major phase of galena and five had an amorphous carbon base. Of the remaining samples, five had different major phases (calcite, cuprite, goethite, silicon and talc) and two were unknown. In the ancient kohl, lead carbonate, hydroxide was found in two samples, chloride found in two of the three, lead sulphide was the major component in two samples and the content of the final sample could not be determined. The SEM data was very similar to what has been stated above, but with the exception that Mo was found in one sample (Hardy et al., 2006). A more recent study from 2017 also employed XRD and EDX (Energy Dispersive X-ray spectroscopy) as well as ICP-MS (Inductively Coupled Plasma – Mass spectroscopy) to analyse Omani-made kohl. Again, XRD revealed galena in four samples, hematite and geotith one sample each. However, eight samples had an amorphous carbon base. EDX showed one sample with lower carbon levels (42%) but for the remaining amorphous carbon samples the carbon level was between 74%-88%, basically being soot (El- Shafey & Al-Kitani, 2017). Similar to XRD, XRF (X-Ray fluorescence) has been used to analyse Egyptian market samples, where 85-89% lead along with 9-13% sulphur were the main components (Bassal et al., 2013). This is also consistent with another study (Ullah PH et al., 2010).

Another method that can be seen in the literature is Raman spectroscopy. In a comprehensive study 133 grey and black samples were collected from the Casa Bacco House in Pompei and were analysed with micro Raman. Most of these containers’ hade a base of calcium carbonate, carbon, silica or iron. However, they acknowledged that the events in Pompei at 79 AD can have altered the content (Gamberini et al. 2008).

El-Shafey & Al-Kitani (2017) also employed ICP-MS for their analysis, for trace metals.

They prepared their samples, as follows: 2g sample was mixed with 10 mL nitric acid (69%) before it was heated to 80℃ and held at this temperature until no brown fumes were visible.

Then 3 mL perchloric acid was added and the sample was again heated to 80℃, followed by the addition of more perchloric acid until white fumes were visible. Then the samples were dissolved in 1 M nitric acid before filtration and washing with MilliQ water. The sample was then diluted to 100 mL. The authors reported a relative standard deviation for the digestion and analysis of 5%. A version of this digestion method was first used to digest kohl and various cosmetics available in Nigeria (Nnorom et al., 2005). Another digestion method that is available in the literature is dry ashing in a porcelain crucible at 550℃ for a few hours and then digestion with 1 M nitric acid, followed by filtration (Ullah H et al., 2017). This

digestion was done before FAAS analysis.

Based on these previous studies, SEM/SEM-EDS, XRD and ICP-AES were chosen to analyse

the samples in this project. The acid digestion method described above has been used in

various studies and thus was applied in this project in order to transform the solid samples

into a liquid phase for ICP-AES measurements. ICP-AES was used instead of FAAS and ICP-

MS that have been reported in previous studies. The application of these three methods should

prove to be viable to determine the main components with SEM-EDS verified by ICP-AES

that also can provide trace elements. SEM will provide the topography and XRD can give an

insight about the structure. Raman was not used due to lack of equipment, ICP-MS was not

used either, due to malfunctioning equipment. However, these three methods combined can

provide a comprehensive picture of the content of these kohl pots, see below for a brief

theoretical introduction to these methods.

7 SEM/SEM-EDS

SEM uses an electron beam that once it hits the surface of the sample some electrons scatter back (BSE – back scatter electrons). Another type of electrons that are detected are the secondary electrons (SE). The BSE and SE are the two main types of electrons that are used to create a grayscale image of the sample’s topography with very high magnification. When the electron beam hits an atom, the atom can absorb energy which excites an electron to a higher orbital. When the excited electron falls back to its ground state X-rays are emitted and detected. Bombarding a sample with many electrons provides a “fingerprint” of the electron configuration in the sample, which is then compared to a reference database. This technique can be used in two ways: either focusing the electron beam on one spot or to scan a larger area.

XRD

X-ray diffraction (XRD) is a based on a 106-year-old idea cultivated by Max von Laue that uses X-ray interference. More precisely it uses a radiation source that hits atoms in different planes of the crystal. As the X-ray scatters, destructive interference will cancel out most waves whereas where positive interference occurs, a signal will hit the detector. If the angle of the radiation source is changed, a diffraction pattern is formed.

The diffraction pattern that is formed according to Braggs Law is: 2dsinθ = nλ, where n is any integer (see figure 3), θ the angle of the radiation, that is changing. From the diffraction pattern an electron density map is created. This pattern is then used to create a “fingerprint”, different crystals form different patterns, that is then compared to reference “fingerprints” in a database of know samples to obtain the crystalline structure (Eckert, 2012).

Figure 3 A schematic view of the principles behind XRD. X-rays are emitted from the radiation source and are scattered on the atoms in the crystal plane, in the right conditions, as seen in the picture, constructive interference will give a signal in the detector.

ICP-AES

ICP works generally by sucking up a small amount of sample which leads into an nebulizer

where a fraction of the starting amount is transformed into tiny droplets. This aerosol then

reaches the argon plasma (up to 10 000K) that atomises the sample. These atoms then absorb

energy and thus get excited to a higher energy level. The excited atoms then fall back to a

lower state and emit energy as light, that then passes through a monochromator. The light’s

8

wavelength is highly specific for each element (a width of around 10 ^-4 nm), making it possible

to capture the light with a detector array and to identify the elements in the sample (Harris,

2010).

9

Experimental

The experimental part of this project is outlined chronologically below.

Sampling

The kohl pots (figure 4) were originally collected near the Egyptian/Sudanese border during SJN (Söderberghon & Troy, 1996) and brought to Gustavianum, Uppsala University Museum where the kohl pots have been stored since the 1970s. The kohl was collected from the kohl pots with a stainless-steel spatula that was cleaned with 95% ethanol between samples. The amount of sample collected was determined by the amount of kohl in the individual pots (about 8 to 30 mg). In one of the pots there was no visible kohl and a sample was collected with a cotton swab. During the sampling, four blank samples were collected, three with no sample and one with an unused cotton swab, see appendix I – samples for individual archaeological notations on the samples. The samples were stored in 15 mL Falcone tubes.

Figure 4. The kohl pots that the samples were collected from. The kohl pots in the top row (A) are from site 185. The pot in the bottom left (B) is from site 183. Bottom middle pot (C) is from site 170. The last one, (D) was the one swabbed with a cotton swab, also from site 185. The pen is used as a scale.

SEM/SEM-EDS

For this experiment the eight samples were placed on a sticky carbon film on an aluminium stud in the SEM-EDS TM-1000-m-DeX´s chamber (see figure 5). To keep track of which sample was which, scratches were made in the carbon film to create sections; an additional scratch was made at the first sample’s location to mark the beginning of sample order.

Figure 5 the 8 samples on a sticky carbon film on top of the aluminium

stub that was placed in the TM-1000-m-DeX SEM-EDS.

10 When eight samples are analysed at the same time like this, it saves a lot of time since the vacuum only needs to be established and broken once during the analysis. However, this caused some issues when retrieving the samples since the small samples are close together on a sticky surface, making them hard to collect without mixing them up.

XRD

The equipment used in this analysis was Bruker D8 SMART Apex-II were the sample was mounted in a capillary tube and the openings where melted to seal the capillary (see figure 6).

The tube was then mounted and slowly turned (2 seconds/ degree) and both the detector and radiation source were stationary.

Figure 6 The Bruker D8 SMART Apex-II XRD that was used during this experiment. The samples where not cooled with liquid nitrogen.

A drawback of placing the sample in a capillary tube is that the sample is then trapped in the tube. If further analysis is to be carried out, the sample must be removed from the tube. This can be done by crushing it, causing glass to mix with the sample that needs to be removed or the tube can be digested along with the sample, again causing contaminations. This turns a non-destructive method into a semi-destructive method.

Sample preparation for ICP-AES

For ICP to function, it generally requires a liquid sample, which means that the samples,

weighing 7 to 16 mg for normal samples and 4.58 g for the swabbed sample (see appendix I

for all weights), need to be digested before analysis (all weighing was done using the same

Mettler AT200 analytical scale). The digestion was made with 1-2 mL nitric acid (Merck

Millipore, Emsure ® ISO EMD analytical grade, HNO 3 65%) at 80℃ until the brownish

vapor dissipated and the sample was almost dry. The sample was then left to cool down

before 1-2 mL perchloric acid (Merck Millipore, Suprapur ® analytical grade, 70% HClO 4 )

was added and the sample was again heated until white fumes developed, which indicated the

11 completion of the digestion. The remaining perchloric acid was evaporated until the sample was almost dry. Then the digested sample was dissolved in 5-7 mL 1M nitric acid diluted in MilliQ water (Millipore’s MilliQ UF system ZFMQ 230 U4)) to dissolve more of the sample, and they were then put in an ultrasonic bath. However, one blank (nr. 3) tipped over and got flooded with water, at which point the ultrasonication was aborted and blank number 3 was excluded from the results. The samples were then filtrated (STORA Filter Products Munktell Filter Paper, 7 cm 00M) into 10 mL measurement flasks, washed with concentrated nitric acid and diluted to the mark with MilliQ water. From the 10 mL flasks an aliquot of about 2 g was accurately weighed into 25 mL flasks and again diluted to the mark with MilliQ water. The estimated sample concentration in the 25 mL flask was in the order of 100 ppm. (see figure 7).

Due to the sensitivity of ICP-AES, this was a good sample concentration to measure trace elements, however it is too concentrated to measure lead. Another dilution was made where about 3.5g was accurately weighed into a 25 mL flask before dilution to the mark with MilliQ water. This dilution brought the estimated concentration of the sample down to have a

concentration in the order of 10-ppm, which is within the linear range of ICP-AES. This was mainly used for lead analysis.

Figure 7 A schematic view of the sample preparation. The ultra-sonification is not in the picture since it was aborted.

For the sample that was collected with a cotton swab additional steps were taken. This was done since perchloric acid can react violently with organic material. The tip of the cotton swab was cut off and then dissolved as above with the addition of concentrated hydrogen peroxide (AnalaR Normapur analytical grade 30% H 2 O 2 ), which completely dissolved the cotton. When the cotton was dissolved the small piece of plastic core in the swab was rinsed off with concentrated nitric acid before being removed. Then perchloric acid was added to complete the digestion. This sample was diluted in the same way as described above (se appendix 1 weighings).

ICP-AES

The equipment used in this experiment was a Spectro Sirrous ICP-AES that utilized a

modified Lichte nebulizer. The Lischte nebuliser transforms the sample to small droplets that are then passed through a spray chamber where only the smallest droplets are carried to the plasma. In this method about 1% of the sample, has the appropriate droplet size, about 8 μm, to be carried into the plasma for analysis. The line width of the emissions is about 25*10 ^-4 nm.

Instrumental settings that were used: 1 400 W, argon plasma flow 14 L/min, auxiliary gas flow 0.9 L/min, nebuliser gas flow 0.9 L/min, sample flow rate 2 mL/min.

As a reference, a calibration curve was made with 0, 0.5, 1, 5 and 10 ppm concentration of the

elements that were of interest. The standard used for the calibration curve was Inorganic

12 Ventures 50.00mg/L analytical grade Multi Standard (d=1.032g/mL), that contained all

elements of interest except sulphur. The sulphur was added using Spectrascans analytical

grade sulphur standard (Sulphur 1.009 µg/mL ± 3μg/mL d=1.000 g/mL). This calibration

curve was then used to set up a method for the ICP-AES, where the standard emission lines

were adjusted to match the actual emission lines that were detected in the calibration curve.

13

Results

The results will be presented and summarized, method by method. See appendix II for raw data.

SEM/SEM-EDS

From the SEM-EDS data, a clear trend could be seen with high concentrations of lead in

almost all samples (see table 1 and figure 8), with an average of 81.3% (weight) lead and

93.1% being the highest measurement. When comparing the lead weight to sulphur, 10.8% on

average, when converting it, the molaric ratio lead/sulphur is close to 1/1 in many samples,

with an average of 1.2 (in favour of lead). Sample 185/30:1 was the only sample containing a

high concentration of silicon. Another sample to consider is 170/14:2, which was the only

sample containing high values of zinc. The zinc was clearly divided into zinc containing

crystals and crystals containing no zinc. These crystals had a distinctly different colour when

viewed with SEM (see figure 9). Other elements, such as iron, arsenic, calcium and chlorine

were also detected, but in much smaller quantities. In nearly all samples aluminium was

detected, however the samples were analysed on an aluminium plate, making it very difficult

to tell if the aluminium was a part of the analyte or background noise. The aluminium data

was only included for samples that clearly had a different composition.

14

Table 1. Summarized results from the SEM-EDS, See Appendix II for the raw data. The average Pb:S molar ratio is calculated for the total lead/sulphur values. When calculating the total average values, the average values from one sample was used. This was done so that one sample would not impact the total disproportionately. Magnification is abbreviated Mag.

Figure 8 a visualisation of the average concentrations (in percent) measured with SEM-EDS in table 1.

Sample Mag. Pb (%)

S (%) Fe

(%) As (%)

Zn (%)

Ca (%)

Si (%)

Al (%)

Cl (%)

Pb:S molar ratio 185/22:1 1 500 86.9 11.7 ND ND 1.5 ND ND ND ND 1.1 185/22:1 5 000 93.1 6.9 ND ND ND ND ND ND ND 2.1 185/22:1 6 000 83.6 16.4 ND ND ND ND ND ND ND 0.8 Average

185/22:1

87.7 11.7 ND ND 0.5 ND ND ND ND 1.2 185/269:1 1 500 79.2 11.5 ND ND ND 5.6 3.7 ND ND 1.1 185/269:1 2 000 86.9 13.1 ND ND ND ND ND ND ND 1.0 Average

185/269:1

83.1 12.3 ND ND ND 2.8 1.9 ND ND 1.0 185/442:2 1 000 62.9 4.0 ND ND ND 7.6 22.2 3.4 ND 2.4 185/442:2 1 500 61.5 1.7 ND ND ND 3.5 25.6 7.8 ND 5.6 Average

185/442:2

62.2 2.9 ND ND ND 5.6 23.9 5.6 ND 3.4 185/30:1 3 000 90.6 9.4 ND ND ND ND ND ND ND 1.5 185/274:1 3 000 88.7 11.3 ND ND ND ND ND ND ND 1.2 185/73:2 2 000 87.9 8.4 3.7 ND ND ND ND ND ND 1.6 183/54:1 3 000 86.9 12.8 ND 0.3 ND ND ND ND ND 1.0 170/14:2 3 000 82.1 12.7 ND 0.1 ND 4.2 0.8 ND ND 1.0 170/14:2

(dark crystal)

4 000 83.1 13.2 ND 0.1 ND 2.4 1.2 ND ND 0.98 170/14:2

(light crystal)

4 000 32.4 26.8 ND ND 29.6 ND 2.2 ND 1.5 0.18 Average

170/14:2

65.9 17.6 ND 0.1 9.9 2.2 1.4 ND 0.5 0.58 Total

average

81.6 10.8 0.5 ND 1.3 1.3 3.4 0.7 0.1 1.2

15 In addition to the SEM-EDS data, several photographs were taken under the SEM to compare the topography of the samples. When using the SEM most of the samples appeared to be visually similar with a few exceptions, see figure 9 for some example photos and appendix II for all photos.

Figure 9 Four of the pictures taken with SEM, see appendix II for all of them. The two to the left, (A and C) are from 17/22:01 with 2 000- and 3 000-times magnification respectively. B shows the different colours of crystals found in 170/14:2 (2 000 times magnification). D shows the slightly different structure of 185/30:1 (1 000 times magnification).

XRD

The crystal structure was analysed on three samples, one sample (185/269:1) was chosen at

random, one sample (185/30:1) was chosen since it had a visibly different colour and the final

sample (170/14:2) was the one whit light and dark crystals which stood out in the SEM-EDS

measurements (see figure 10-14). From the data, Pb 2 O(SO 4 ) was the major phase although

PbS was still present in two of the samples. In the, possibly contaminated, sample SiO 2 was

the major phase.

16

Figure 10. Sample 185/269:1s (random sample) diffractogram: from it Pb2O(SO4) and PbS was identified, with the former as a major phase.

Figure 11 sample 170/14:2s (light and dark crystals) diffractogram: from it Pb2O(SO4) and PbS was identified, with the former as a major phase

Figure 12 reference diffractogram for PbS and Pb2O(SO4) respectively -5000

0 5000 10000 15000

0 10 20 30 40 50 60

185/269:1

-5000 0 5000 10000 15000

0 10 20 30 40 50 60

170/14:2

17

Figure 13 Sample 185/30:1s diffractogram: This is the sample that was possibly contaminated. Here SiO 2 was identified as a major phase

Figure 14 reference diffractogram for SiO 2 .

Digestion

The efficiency of the digestion, prior to ICP-AES, was somewhat varied. In initial testing the sample 185/22:1, the digestion was efficient, with all but a tiny speck being dissolved, but this did not work on all samples particularly 185/30:1 (the, possibly contaminated, sample) and 185/442:2 with 185/274:1 as intermediate (the middle one), see figure 15. The sample that was collected with a cotton swab dissolved completely into a clear liquid.

Figure 15 The three worst results after digestion. From the left the samples are 185/30:1, 185/274:1 and 185/442:2. The remaining samples contained much less precipitate.

-50000 0 50000 100000 150000 200000 250000 300000 350000

0 10 20 30 40 50 60

185/30:1

-20000 0 20000 40000 60000 80000 100000 120000

0 10 20 30 40 50 60

SiO2-triclinic

18 ICP-AES

When measuring with ICP-AES, two different emission lines were used for all elements, except potassium, which did not have a good second emission line. Table 2 and 3 shows the concentrations obtained from the different emission lines I was advised to use, where 100 ppm and 10 ppm sample refer to the dilution order of the sample. The lower concentration was necessary to measure in the linear range for lead in ICP-AES.

Table 2 Set 1 of the measurements done with ICP-AES, the data is presented as percent of the digested sample. Note that

Sample 185/219:2 was the sample that was collected with a cotton swab, hence the much lower concentrations. Also note

that the Pb to S ratio is far from 1:1

19

Table 3 Set 2 of the measurements done with ICP-AES, the data is presented as percent of the digested sample. Note that Sample 185/219:2 was the sample that was collected with a cotton swab, hence the much lower concentrations, especially in the 10-ppm dilution range. Also note that the Pb to S ratio is far from 1:1

From table 2 and 3, the average concentrations are visually presented in figures 16.

Figure 16 side by side compression for the concentrations (in percent) in the two sets of emission lines that were used for the

ICP-AES. Above are the concentrations measured from the 10 ppm samples and below for the 100 ppm samples.

20 Principal Component Analysis (PCA)

Once the samples had been measured with ICP-AES, the concentrations from emission line 1 in the 100 ppm samples were analysed with PCA (The Unscrambler v. 9.7, CAMO software AS), except for the sample that had been collected with a cotton swab (185/19:2). This was done to see if it was possible to see a difference from the samples in the different grave sites.

This did not produce good results. To obtain a better picture, two samples were excluded, these were the two samples 185/30:1, that was possibly contaminated and 185/442:2, where the digestion did not appear to have worked well, (see fig 17).

From the loading plot (figure 18) Pb, and a cluster of elements (Fe, Ca, Si, Ba, Sr, Al, Ba) are very important for determining the orientation along the x-axis (the 1 ^st component). It appears that Zn, S and As contribute a lot to the position in the y-axis (2 ^nd component).

Figure 18. The loading plot behind figure 17. Pb, Zn, S and As seem to be the individually most important components where the rest of the elements make up a cluster that contributes to the first principal component.

Al Si

Zn

Pb S

As Fe

Ca Ba

Mg

Sr

-0,6 -0,4 -0,2 0 0,2 0,4 0,6 0,8

-0,4 -0,3 -0,2 -0,1 0 0,1 0,2

Loading plot

Figure 17 Scoring plot from spectral lines 1, whit three samples excluded (185/19:2, 185/30:1 and 185/442:2) Samples from site 185 are marked red, the sample from site 183 is green and that from site 170 is purple.

185/22:1 185/269:1

185/274:1 185/73:2

183/54:1

170/14:2

-2,5 -2 -1,5 -1 -0,5 0 0,5 1 1,5 2 2,5

-6 -5 -4 -3 -2 -1 0 1 2 3

Score plot based on 100-ppm concentrations

from ICP-AES line 1

21

Discussion

When comparing the results form SEM-EDS, ICP-AES as well as XRD it is easy to spot differences in the results. The results of the SEM-EDS had an average of 81% lead, compared to ICP-AES that yielded an average of 40 to 44% lead, depending on emission line. This is obviously a huge discrepancy. ICP-AES is very accurate on lead concentrations in the 10-ppm range. To account for the discrepancy either the SEM-EDS produces too high values, or the digestion did not work as well as previous studies have reported. Considering that a few samples contained quite a lot precipitate after the digestion I would have to say that the digestion did not work as well as intended, especially in sample 185/30:1, the, possibly contaminated one, after digestion there was quite a bit of solid matter left. Considering that the SEM-EDS showed it to contain about 25% Si and XRD showed a major phase of SiO 2 it is not surprising that it did not dissolve as well with a method to dissolve lead. Also, when looking at the sulphur concentrations from ICP-AES compared to SEM-EDS there is a large discrepancy, 2 and 10 % respectively. This is an even larger discrepancy and can be explained by both an incomplete digestion as well as the fact that sulphur evaporates during the

digestion. Generally, SEM-EDS showed higher values on all elements, except for zinc. When it comes to arsenic, the ICP-AES had an average concentration of 0.01% compared to the results with SEM-EDS, which showed one sample with a 0.3% concentration. It is difficult to say which result is correct, since the As levels measured in SEM-EDS were close to the detection limit. But as As is present in both SEM-EDS and ICP-AES it stands to reason that some samples contain traces of arsenic. Another interesting data from ICP-AES was the amount of aluminium, an average of about 0.1% compared to the SEM-EDS data where aluminium showed up in all samples. Since the samples in SEM-EDS were mounted on an aluminium stud the aluminium concentration could not be trusted. As only small traces of Al were found using ICP-AES, it is important to confirm that the aluminium recorded in SEM- EDS was due to background interference. When comparing the Ca concentrations between the 10 and 100 ppm sample concentrations, the concentrations are higher in the 10 ppm samples.

The reason for this is that the blank concentrations for the 10-ppm sample for Ca were a lot lower than the 100 ppm Ca blank. ICP-AES is very sensitive for Ca and the linear range is very low, but this factor does not account for this disparity. A plausible explanation is probably Ca contamination of the glassware in which the 100-ppm blank sample was prepared.

From the SEM-EDS data most of the lead is in galena (PbS) crystals, however XRD data suggests that Pb 2 O(SO 4 ) is the major phase. This could be due to PbS specks having a coating of Pb 2 O(SO 4 ). This could intern give false data from the XRD, since it does not penetrate the sample enough. Given this I believe that the true major phase is PbS, but that the samples also contain the oxidation product, Pb 2 O(SO 4 ).

Even though the ICP-AES data cannot be considered to be reliable for precise absolute quantification, it can still be used for a precise relative quantification. To take advantage of this a PCA was done to attempt to group the samples that were similar to each other. From PCAs loadings one can argue that the samples from site 185 are quite similar regarding the

“cluster elements” (Fe, Ca, Si, Ba, Sr, Al, Ba), although one sample (185/269:1) has a lower

Pb concentration and a higher concentration of the “cluster elements”. In comparison the

sample from site #183 stands out quite a bit, being the only one containing zinc. This,

combined with sample 170 being the only one in its quadrant suggests that there might have

been small local variation of the contents in the kohl samples. However more samples from

the different sites would be needed to evaluate this.

22 Conclusions

The majority of the content in the kohl that was analysed was lead, with some SEM-EDS measurements showing that over 90% (average of 80%) of its mass being lead. This combined with high levels of sulphur, an average of 10% (SEM-EDS) suggests that the main component of the kohl is PbS. The presence of PbS was confirmed with XRD. However, XRD indicated that Pb 2 O(SO 4 ) was also present. Although good quantification from ICP-AES is difficult due to issues with digestion some trace elements are common, with calcium being the most

abundant (1.70%), followed by zinc (1.5%) and with silicon, magnesium and iron at around 0.33%. Strontium and aluminium concentrations were in the 0.1 % range. The PCA analysis can, unfortunately, not be used to draw conclusions from, since there are too few samples in it. However, with more samples from each sites I believe it can be used to draw conclusions regarding small, local variances in ether the production, storage or the raw materials used in the kohl.

Two samples stood out, one was sample 185/30:1, that had a different colour and all the analyses indicated a different chemical makeup. The XRD showed a clear SiO 2 phase, from SEM-EDS we get 25% silicon. ICP-AES backed this up with 185/30:1 containing 0.125 of the amounts of lead compared with other samples. The other sample that stood out was 185/219:2, which had been collected with a cotton swab; unfortunately, SEM-EDS and XRD measurements were not possible on this sample. When measuring with ICP-AES only 0.06%

lead was detected; this was the weight % from the cotton swab tip. However, if the

concentration in the 10 mL flask is used instead of in the swab, around 0.5% lead was present.

Maybe a different digestion method or the use of ICP-MS would make it possible to provide better data on trace amounts of kohl.

Future aspects

To improve upon this method, I suggest a different digestion strategy. Instead of digesting in an open beaker I would try digestion under pressure. However, when analysing these types of samples, microwave digestion (MWD) which usually uses vessels with a Teflon coating on the inside, can prove difficult to use with lead, since it can strongly adsorb to the vessel, contaminating it for a long time. Using a quart’s vessel with, a steel casing and boiling nitric acid together with perchloride or another strong acid for a longer duration would probably provide a more complete digestion and a higher retention of sulphur, leading to a more reliable ICP-AES quantification. This along with XRD measurements on all samples would be a good improvement. If the same number of samples that were analysed in this study where to be analysed with the same numbers of samples from other sites it could be possible to determine whether or not local influences or materials cause a difference in the composition of the kohl.

Another way to improve upon the results would be repeated determinations. This way one would be able to, more statistically correct, determine the concentrations.

23 With the development of powerful modern-day analytical techniques, such as ICP-MS which allows for determining isotope ratios in an accurate and reliable way, chemistry’s role in archaeology will grow over time. The fact that – barring contamination – a minute amount of sample that can provide a lot of very accurate hard data is proving very useful to study limited quantities of cultural material. If a large-scale isotope analysis would be done on all mines in the area, in order to provide reliable isotope ratio references for the mines, isotope analysis of artefacts could in the future be used to accurately place their origin. This type of analysis is already being done (Shortland, 2006). In time I believe it will become a normal, everyday practice. The digested samples have been saved for future studies, such as isotope analysis. To get a clear, comprehensive picture of the contents of the kohl pots other methods could be applied, for example Raman spectroscopy.

In Gustavianum’s collection there is a kohl pot with a sealed lid, indicating that its content has

not been exposed to the environment, as was not the case with the samples in this study. A

thorough analysis of that kohl pot could prove to be an invaluable reference material in future

studies as well as confirming or providing an indication of the contamination levels of the

result obtained in this study.

24

Thanks to

Jonas Bergquist, Professor of analytical chemistry and neurochemistry UU, associate professor of clinical neuroscience at the Sahlgrenska University Hospital and Gothenburg University – Chemistry supervisor

Emma Hocker, Senior Conservator at Gustavianum, UU– Archaeological supervisor Jean Pettersson, lector in analytical chemistry – Aided with ICP

Vadim Kessler Professor of Inorganic Chemistry and Bionanotechnology, SLU – Aided with XRD and SEM-EDS

Marit Andersson, lector in analytical chemistry UU – Subject examiner

Helena Grennberg, Professor of organic chemistry UU – Examinator

Patricia Kindström, Chemistry bachelor student, Opponent

25

References

Bassal N, Mahmoud HH, Fayez-Hassan M. 2013. Elemental Composition Study of Kohl Samples. 8.

CAMO Software AS (2007), The Unscrabler V.9.7

Crawford C, Krebs DL. 2013. Handbook of Evolutionary Psychology: Ideas, Issues, and Applications. Psychology Press

Eckert M. 2012. Max von Laue and the discovery of X-ray diffraction in 1912. Annalen der Physik 524: A83–A85.

El-Shafey E-SI, Al-Kitani BSH. 2017. Comparative chemical analysis of some traditional Omani-made kohl. Toxicological & Environmental Chemistry 99: 233–251.

Genge H. 1979. Nordsyrisch-südanatolische Reliefs: eine archäologisch-historische Untersuchung, Datierung und Bestimmung. 1: Text. Munksgaard, København.

Hardy AD, Sutherland HH, Vaishnav R. 2002. A study of the composition of some eye cosmetics (kohls) used in the United Arab Emirates. Journal of Ethnopharmacology 80: 137–

145.

Hardy AD, Walton RI, Vaishnay R, Myers KA, Power MR, Pirrie D. 2006. Chapter 5

Egyptian eye cosmetics (“Kohls”): Past and present. Physical Techniques in the Study of Art, Archaeology and Cultural Heritage, pp. 173–203. Elsevier,

Harris DC. 2010. Quantitative chemical analysis, 8th ed. W.H. Freeman and Co, New York.

Ko D. 2005. Cinderella’s Sisters: A Revisionist History of Footbinding. University of California Press

Parry C, Eaton J. 1991. Kohl: A Lead-Hazardous Eye Makeup from the Third World to the First World. Environmental Health Perspectives 94: 121.

Scott DA. 2016. A review of ancient Egyptian pigments and cosmetics. Studies in Conservation 61: 185–202.

Shortland AJ. 2006. Application of lead isotope analysis to a wide range of late bronze age egyptian materials. Archaeometry 48: 657–669.

Sulaiman bin Ash'ath (tralsated by Yaser Qadhi) (2008). lmâm Hâfiz Abu Dawud Stos-Gale ZA, Gale NH. 1981. Sources of Galena, lead and silver in predynastic Egypt.

Revue d’Archéométrie 1: 285–296.

Säve-Söderberghon and Troy, 1996. New Kingdom Pharaonic Sites The finds and sites. The

University of Chicago Press 55:2.

26 Tapsoba I, Arbault S, Walter P, Amatore C. 2010. Finding Out Egyptian Gods’ Secret Using Analytical Chemistry: Biomedical Properties of Egyptian Black Makeup Revealed by Amperometry at Single Cells. Analytical Chemistry 82: 457–460.

Tiffany-Castiglioni E, Barhoumi R, Mouneimne Y. 2012. Kohl and surma eye cosmetics as significant sources of lead (Pb) exposure. Journal of Local and Global Health Science 1.

Ullah H, Noreen S, Fozia, Rehman A, Waseem A, Zubair S, Adnan M, Ahmad I. 2017.

Comparative study of heavy metals content in cosmetic products of different countries marketed in Khyber Pakhtunkhwa, Pakistan. Arabian Journal of Chemistry 10: 10–18.

Ullah PH, Mahmood ZA, Sualeh M, Zoha S. 2010. Studies on the chemical composition of kohl stone by x-ray diffractometer. Pak J Pharm Sci 6.

Waddington, Ray (2002), The Karen People. The Peoples of the World Foundation. Retrieved October 24, 2018, from The Peoples of the World Foundation.

<http://www.peoplesoftheworld.org/text?people=Karen>

27

Appendix I Samples & Sample Preparation

Table A.1.1 The spatula was cleaned, with ethanol between each sample. Legend: XXX/XXX:XX Site/Grave: find Sampling date: 2018-09-06. Sample number is the notation used on the samples during the labwork.

Sample number

Arkeological notation/content

Comments

B1 air blank

B2 air blank

B3 air blank

B4 swab blank

S1 17/22:01 Kohl pot had lid – Black Sample

S2 185/269:1 Black sample

S3 185/442:2 Black sample

S4 185/30:1 Brown/yellowish colour, could contain fragments of the pot or other contaminations.

S5 185/274:1 Black sample

S6 185/73:2 Black sample

S7 183/54:1 Black sample

S8 170/14:2 Kohl pot had lid – Black sample

S9 185/219:2 Collected with swab

Tabell A.1.2 Sample weight used in the digestion as well as the weight transfer with a pipet (assumed density =1). All weighing where done on the same analytical scale.

Sample Sample mass

(mg)

Transferred amount. 1st dilution (mL)

Transferred amount. 2nd dilution (mL)

Blank 1 ND 2.1653 3.2267

Blank 2 ND 2.0805 3.5042

Blank 3 ND 1.9557 3.6804

Blank 4 (cotton swab) 0.0473 1.9523 3.4240

185/22:1 18.5 2.0852 3.1651

185/269:1 11.9 1.9958 3.2703

185/442:2 7.0 2.0727 3.4917

185/30:1 16.7 1.9670 3.5469

185/274:1 13.9 2.0076 3.6350

185/73:2 9.20 2.0166 3.6190

183/52:1 16.2 1.9815 3.7514

170/14:2 10.5 2.0115 3.6812

185/219:2 45.8 1.9653 3.7589

28

Appendix II – Sem & SEM-EDS data

Sample 17/22:01

Figure A.2.1. 10 000 times magnification of sample 17/22:01

Figure A.2.2 6 000 times magnification, 17/22:01

Figure A.2.3 3 000 times magnification

29

Figure A.2.4 500 times magnification, area scan

Table A.1.1 Auto identification yielded the following results for 17/22:01 500 magnification

Element Weight % Sulphur 11.7

Zinc 1.5

Lead 86.9

Figure A.2.5 5 000 times magnification, sample 1, 2nd area scan

Table A.1.2 Auto identification yielded the following results for 17/22:01, 2nd scan, 5 000 magnification.

Element Weight % Sulphur 6.9

Lead 93.1

Figure A.2.6 6 000 times magnification, 17/22:01, 3rd scan

30

Table A.2.3 Auto identification yielded the following results for 17/22:01,3rd scan, 6 000 magnification.

Element Weight %

Sulphur 16.4

Lead 83.6

Sample 185/269:1

Figure A.2.7 Sample 185/269:1 at 500 times magnification

Figure A.2.8 Sample 185/269:1 at 2000 times magnification

Figure A.2.9 sample185/269:1, 500 times magnification Table A.2.4 185/269:1, 1 500 times magnification

Element Weight % Silicon 3.7 Sulphur 11.5 Calcium 5.6

Lead 79.2

31

Figure A.2.11 Sample 185/269:1, 2 000

Table A.2.5 Sample 185/269:1, 2 000 times magnification

Element Weight % Sulphur 13.1

Lead 86.9

Sample 185/442:2

Figure A.2.12 sample 185/442:2, at 1 000 times magnification

Figure A.2.13 Sample 185/442:2, 1 000 times magnification Table A.2.6 Sample 185/442:2, 1 000 times magnification

Element Weight % Aluminium 3.4

Silicon 22.2

Sulphur 4.0

Calcium 7.6

Lead 62.9

32

Figure A.2.14 sample 185/442:2, 1 500 times magnification Table A.2.73 sample 185/442:2, 1 500 times magnification

Element Weight % Aluminium 7.8

Silicon 25.6

Sulphur 1.7

Calcium 3.5

Lead 61.5

Sample 185/30:1

Figure A.2.15 Sample 185/30:1 at 3 000 times magnification

Figure A.2.16 sample 185/30:1, 3 000 times magnification Tabell 1 A.2.8 Sample 185/30:1, 3 000 times magnification

Element Weight % Sulphur 9.4

Lead 90.6

33 Sample 185/274:1

Figure A.2.17 sample 185/274:1 at 3 000 times magnification

Figure A.2.18. Sample 185/274:1, 3 000 times magnification Table A.2.9 sample185/274:1, 3 000 times magnification

Element Weight %

Sulphur 11.3

Arsenic ND

Lead 88.7

Sample 185/73:2

Figure 2 sample 185/73:2 at 2 000 times magnification

34

Figure A.2.20 Sample 185/73:2 2 000 times magnification Table A.2.104 sample 185/73:2, 2000 times magnification

Element Weight % Sulphur 8.4

Iron 3.7

Lead 87.9

Sample 183/54:1

Figure A.2.21 sample 183/54:1 at 500 times magnification

Figure A.2.22 Sample 183/54:1, 3 000 times magnification

35

Table A.2.11 Sample 183/54:1, 3 000 times magnification

Element Weight % Sulphur 12.8 Arsenic 0.3

Lead 86.9

Sample 170/14:2

Figure A.2.23 Sample 170/14:2 at 2000 times magnifcation

Figure 34 sample 170/14:2, 3 000 times magnification Tabell A.2.12 sample 170/14:2, 3 000 times magnifcation

Element Weight %

Silicon 0.8

Sulphur 12.7

Calcium 4.2

Arsenic 0.1

Lead 82.1

36

Figure 4 Sample 170/14:2, light crystall at 4 000 times magnification Table A.2.13 Sample 170/14:2, light crystall at 4 000 times magnifcation

Element Weight % Silicon 1.2 Sulphur 13.2 Calcium 2.4 Arsenic 0.1

Lead 83.1

Figure A.2.26 Sample 170/14:2, dark crystal at 4 000 times magnification Table A.2.14 Sample 170/14:2, dark crystal at 4 000 times magnification

Element Weight % Silicon 2.2 Sulphur 26.8 Chlorine 1.5 Calcium 5.7

Zinc 29.6

Lead 34.2