A Yellow-Green Vase with Handles Attributed to the Hongzhi Period: Results of Portable XRF Analysis

Abstract Portable X-ray fluorescence (pXRF) analysis of the yellow ground and green area of a handled vase attributed to the Hongzhi period (1488–1505) revealed no reliable markers of modern manufacturing technology after spectral artefacts associated with the high-lead matrix had been excluded. Both glazes are high-lead compositions (Pb: 49.40 and 52.03 wt.% as reported by the instrument in elemental mode). The green coloration is copper-based (Cu: 4,280 ppm). Antimony, zirconium, vanadium, and titanium were not detected. These findings exclude lead-antimonate yellows associated with the Qing dynasty and early Republican-period imitations, as well as industrial zircon- and rutile-based pigments. The readings for Cd, Sn, S, As, Zn, Se, Mo, Nb, Hf, and W are interpreted as known artefacts of portable XRF analysis, including pulse pile-up peaks generated by lead, overlap of Pb Lα and As Kα lines, and interference between Pb M and S Kα lines. Two issues remain unresolved: the unexpectedly low iron content in the yellow glaze (Fe: 1,730 ppm, compared with published values of 1.81–4.87 wt.% Fe₂O₃ for Ming yellow glazes) and the chromium reading (approximately 1,600 ppm). A programme of further investigation is proposed. The analytical results are consistent with the attribution but do not independently establish it. Keywords: Ming dynasty; Hongzhi period; Jingdezhen; lead yellow glaze; copper green glaze; portable X-ray fluorescence; pulse pile-up; spectral interference; attribution. 1. Introduction Ming imperial yellow is a low-melting, high-lead glaze coloured with iron. The Hongzhi period is regarded as the peak of this technological tradition, particularly in the production of the characteristic so-called chicken-fat yellow (娇黄). Colourant. According to EDXRF, Raman spectroscopy, and XPS analyses of eleven samples from the imperial kilns of Jingdezhen, the colourant in Ming yellow glaze is Fe₂O₃ dissolved in a lead-glass matrix, with concentrations ranging from 1.81 to 4.87 wt.% (Hao et al., 2024). Historical Ming recipes used lead, quartz, and ochre, or what was described as "pure iron red". Layer structure. A transparent high-temperature layer, 150–450 μm thick and containing numerous bubbles, separates the coloured lead glaze from the body of imperial yellow porcelain. In architectural tiles, by contrast, the lead glaze was applied directly to the ceramic body (Hou et al., 2022). A slip layer has also been reported in Chenghua-, Hongzhi-, and Zhengde-period examples (Zheng, Wu, Wu, 2008). Later yellow glazes. In early Republican-period imitations, yellow coloration was produced using antimony and tin (the phase Pb₂Fe₀.₅Sb₁.₅O₆.₅; Zheng, Wu, Wu, 2008). From the late Kangxi era onwards, lead-antimonate and lead-tin yellows became part of the glaze palette. Industrial twentieth-century yellows were based on zircon (Zr–Pr, Zr–V), rutile (Ti–Sb–Ni/Cr), cadmium, or chromium compounds. Green glaze. Ming low-temperature green glaze was coloured with copper in a lead-rich matrix. The aim of this study is to compare the pXRF readings obtained from the vase with these technological criteria, first distinguishing genuine elemental signals from artefacts associated with a high-lead matrix. 2. Object and Provenance The object under examination is a porcelain vase with handles, covered with a yellow lead glaze and green patches. Its proposed attribution is Jingdezhen, Hongzhi period (1488–1505). In the research register it is designated "Vase 3"; in the author's archive its reference is DRN-MING-YG-2025-HZ. Dimensions, form, and marks: [to be completed]. Provenance (collection history, acquisition dates, publications, and auction records): [to be completed with documentary references]. Documented provenance constitutes evidence independent of the XRF results and should be supported by verifiable sources in the final publication. Surface condition: iridescence is visible on the yellow glaze. Close-up photographs show an ochre-yellow ground, while the green area is olive-green with dark pinpoint inclusions. 3. Materials and Methods Measurements were performed in situ on 3 November 2026 using a portable XRF analyser (serial number 940487), without sampling or removal of material. Parameter Value Analytical mode Geochem (3-Beam) Extra Measurement time 30 seconds per spot Measurement point 1 (Daily ID 4, 13:57:13) Green area, Sample ID 3, "Vase Green" Measurement point 2 (Daily ID 5, 13:58:13) Yellow ground, Sample ID 3, "Vase Yellow" The reported values are elemental concentrations calculated by the instrument's built-in algorithm. Measurement uncertainties are reported as ±3σ. No calibration against high-lead reference standards was performed. The Geochem mode is based on a light silicate matrix model and is not designed for glasses containing approximately 50% lead. Consequently, absolute concentration values are interpreted as semi-quantitative. The reliability of each elemental reading was assessed by examining peak positions in the spectra and comparing the results with other high-lead glazes analysed during the same measurement series, including a yellow vase designated CHENGHUA-YEL-001 (Pb: 48.97%). 4. Results Both measurement points correspond to high-lead glazes. The green area differs from the yellow ground primarily in its copper content; most other readings are similar. Table 1. Instrument readings (ppm unless otherwise indicated; ±3σ) and assessment of reliability Element Green area (ID 4) Yellow ground (ID 5) Assessment Pb 49.40% ± 0.16 52.03% ± 0.18 Genuine Si 14.98% ± 0.14 13.39% ± 0.15 Genuine (absolute values inaccurate) Cu 4,280 ± 260 ND (<39) Genuine K 5,960 ± 390 4,720 ± 410 Genuine, probably underestimated Al 4,100 ± 1,300 4,200 ± 1,300 Genuine, inaccurate Fe 1,500 ± 200 1,730 ± 210 Genuine, probably underestimated (Section 7) P 670 ± 160 800 ± 170 Genuine Mn 260 ± 200 320 ± 200 At detection limit Ca 370 ± 270 ND (<540) Low Cr 1,540 ± 250 1,640 ± 260 Unresolved (Section 7) As 3.234% 3.421% Predominantly artefactual (Section 5) S 28.93% 27.92% Artefact Cd 1,420 ± 300 1,450 ± 310 Artefact Sn 960 ± 330 810 ± 330 Artefact Zn 3,670 ± 140 3,840 ± 140 Probably artefactual Se 256 ± 88 307 ± 93 Probably artefactual Nb / Mo / Hf / W 1,274 / 296 / 2,990 / 4,340 1,388 / 333 / 3,130 / 4,420 Artefacts Pr / Nd ND 1,360 ± 880 / 1,900 ± 1,300 Artefacts (uncertainties comparable to measured values) Sb ND (<12) ND (<11) Not detected Zr ND (<3) ND (<3) Not detected Ti ND (<530) ND (<590) Not detected V ND (<66) ND (<67) Not detected Co / Ni ND (<21 / <11) ND (<21 / <12) Not detected Converted to oxide equivalents, the readings correspond to approximately 53% PbO in the green glaze and 56% PbO in the yellow glaze, approximately 0.54% CuO in the green area, and approximately 0.21–0.25% Fe₂O₃. The spectra from both measurement points are dominated by Pb Lα (overlapping with As Kα) and Pb Lβ peaks, with intensities of approximately 17,000–21,000 counts per second, together with a Pb M peak in the low-energy beam at approximately 7,500–8,500 counts per second. 5. Spectral Artefacts Associated with the High-Lead Matrix All apparently modern elemental readings in Table 1, with the exception of chromium, can be explained by known artefacts of portable XRF analysis in materials containing approximately 50% lead. 5.1. Lead Pulse Pile-Up Peaks: Cd and Sn At high count rates, two photons originating from lead may be registered as a single event with their combined energy. Methodological literature describes the following coincidences: Pb Lα + Pb Lα = 21.0 keV (Pd Kα), Pb Lα + Pb Lβ = 23.1 keV (Cd Kα at 23.2 keV), and Pb Lβ + Pb Lβ = 25.2 keV (Sn Kα). These peaks disappear when the X-ray tube current is reduced (Lee, 2016; WAAC, 2009). The spectra of the high-lead glazes examined in this series display precisely this triplet of peaks, at approximately 21.1, 23.2, and 25.3 keV, with the middle peak having the greatest intensity. Internal confirmation is provided by measurements of three different glazes containing approximately 50% Pb: their Cd readings are almost identical (1,420, 1,450, and 1,420 ppm). By contrast, cadmium was not detected in a glaze containing 7.4% Pb (below 19 ppm). 5.2. False Sulphur Signal The Pb Mα line (2.35 keV) almost coincides with the S Kα line (2.31 keV). A direct methodological analogue is provided by the analysis of high-lead Corning C reference glass, in which portable XRF incorrectly reported approximately 30 wt.% SO₃, while light elements were underestimated. The sulphur readings of 27.9–28.9% obtained from the vase are consistent with the same effect. Some sulphur may nevertheless be genuinely present: lead sulphate has been described as a corrosion product of lead-glazed ceramics (Duan Hongying et al., 2013). 5.3. Arsenic The Pb Lα line (10.55 keV) and As Kα line (10.54 keV) cannot be distinguished spectrally. The instrument manufacturer compensates for the lead contribution using the Pb Lβ line; at high lead concentrations, any inaccuracy in this model can produce a false arsenic signal (Rowe et al., 2015). Within the present measurement series, the As/Pb ratio is identical in three different glazes containing approximately 50% Pb: 0.065 (green), 0.066 (yellow), and 0.066 (CHENGHUA-YEL-001). This strongly suggests an artefact proportional to the lead content. For comparison, low-lead glazes from the same series have As/Pb ratios of 0.2–0.26. 5.4. Other Lead-Related Artefacts Zn, Se, Mo, Nb, Hf, W, and Pr/Nd readings are consistently reproduced in all three glazes containing approximately 50% Pb and are explained by escape and sum peaks associated with Pb and As. The presence of hafnium in the absence of detectable zirconium (below 3 ppm) is physically implausible for a genuine material composition. Research on lead-rich glass tesserae has demonstrated that por