SURVEY OF COSMETICS FOR SEVEN INORGANIC ELEMENTS 137 Table XVIII. Blush Summary (n = 30) Results (mg/kg) As Cd* Cr Co Pb Hg Ni Maximum 1.3 1.2 (0.12) 400 7.5 14 0.010 23 Minimum NF NF TR NF TR NF TR Mean 0.51 0.056 NF 23 2.0 3.9 0.0027 8.0 Median 0.46 NF 6.5 1.5 3.3 NF 7.3 NF: Not found, or less than detection limit TR: Trace, or greater than detection limit, but less than quanti- tation level. *The signifi cantly high value of 1.2 mg/kg for Cd from one sample was removed for the calculated values in parentheses. Table XIX. Lipstick Summary (n = 30) Results (mg/kg) As Cd Cr Co Pb Hg Ni Maximum 0.49 NF 19 3.9 3.4 0.010 7.6 Minimum NF NF 0.18 NF TR NF 0.10 Mean 0.14 NF 2.8 0.86 1.0 NF 2.6 Median 0.13 NF 1.5 0.45 0.82 NF 1.8 NF: Not found, or less than detection limit TR: Trace, or greater than detection limit, but less than quanti- tation level. • Include one organic oil reference material (Conostan oil for non-mercury metals and SPEX oil for mercury) in each digestion batch. • Include one inorganic reference material (NIST 2702) in each digestion batch. • Prepare each sample in duplicate. ICP-MS DETERMINATION OF ARSENIC, CADMIUM, CHROMIUM, COBALT, LEAD, AND NICKEL The ICP-MS apparatus was set up to include the following isotopes: • Elements of interest: 52,53 Cr, 60,62 Ni, 75 As, 59 Co, 111,114 Cd, 206,207,208 Pb • Internal standards: 45 Sc, 115 In, 195 Pt Initial calibrations used the 8 calibration standards from Table I. Continuing calibration verifi cation (CCV) standards were analyzed after every 15 analytical solutions. The manu- facturer’s software was used to compute the linear regression and the digital to analog response ratio. The sum of lead isotopes was used to account for isotopic variation be- tween samples and standards. The isotopes bolded above were used for quantitation and all of the isotopes were used for conformation and interference assessment. Prepared sam- ples were analyzed using a dilution of 1/10 with the diluting solution. Analytical solu- tions were analyzed relative to calibration standards using internal standardization and linear regression (y = ax + blank). No analytical solutions had internal standard signals that differed by more than 40% relative to the calibration blank.

JOURNAL OF COSMETIC SCIENCE 138 Table XX. Lotion Summary (n = 5) Results (mg/kg) As Cd Cr Co Pb Hg Ni Maximum 0.19 NF TR TR 0.037 TR 0.37 Minimum 0.079 NF NF NF NF NF 0.040 Mean 0.12 NF NF NF 0.010 NF 0.11 Median 0.10 NF NF NF NF NF 0.053 NF: Not found, or less than detection limit TR: Trace, or greater than detection limit, but less than quanti- tation level. Table XXI. Premium Lotion Summary (n = 5) Results (mg/kg) As Cd Cr Co Pb Hg Ni Maximum 0.20 NF TR 0.11 0.083 NF TR Minimum TR NF NF NF NF NF TR Mean 0.14 NF 0.084 0.027 0.019 NF 0.062 Median 0.12 NF 0.084 NF NF NF 0.053 NF: Not found, or less than detection limit TR: Trace, or greater than detection limit, but less than quanti- tation level. The correlation coeffi cient for the standard curve was better than 0.997 in all cases. Con- tinuing calibration was verifi ed by analyzing a calibration blank and two QC solutions (secondary check solution and NIST 1643e reference standard) immediately after calibra- tion and after every 15 analytical solutions. The QC solution recovery was between 90% and 110%. One analytical solution from each digestion batch was diluted 1/20 with di- luting solution and compared with the same solution diluted 1/10 to verify absence of matrix effects. Relative percent differences (RPDs) for analytes in solutions at different dilutions were less than 20%. Any analytical solutions with analyte values above the highest standard were additionally diluted to determine the concentrations. CVAFS DETERMINATION OF MERCURY To avoid loss of mercury through volatilization, the analytical solutions were analyzed prior to the evaporation steps used to prepare samples for ICP-MS. As is typical in mer- cury vapor analysis, solutions were fi rst oxidized to ensure all mercury was in a stabilized oxidation state in solution, and then were reduced to form the volatile elemental mercury vapor. Initial calibrations used a blank and 5 calibration standards. CCV standards were ana- lyzed after every 15 analytical solutions. The manufacturer’s software was used to com- pute the linear regression. Oxidized solutions were treated with hydroxylamine hydrochloride to neutralize any remaining oxidant. Solutions were then treated with