At the late-Neolithic walled site of Taosi (Xiangfen, Shanxi, c. 2300–1900 BC), excavations exposed a semi-circular rammed-earth platform, IIFJT1, whose thirteen piers define twelve narrow slots (backsights E1–E12) sighted from a central observation point toward the sunrise horizon. Wu, Chen and Liu (2009) confirmed astronomically that E2 and E12 marked winter- and summer-solstice sunrise at the archaeologically dated epoch (c. 2100 BC), but left open the function of the remaining ten backsights (E1, E3–E11), whose roughly even angular spacing divides the sunrise arc between the solstice slots into roughly equal spatial parts — a geometry whose calendrical meaning (if any) is contested, because even azimuth spacing implies strongly non-uniform intervals between the sun's appearances in successive slots; the anomalous offset of E11 (with E12) from the main arc, suggesting possible multi-period construction; and the epoch of the structure, which alignment-based dating cannot fix because centimeter-scale positional uncertainties translate into centuries of date error. Later work has sharpened but not settled any of these strands: Pankenier, Liu and De Meis (2008) showed that even azimuth spacing implies strongly non-uniform date intervals and proposed that E11–E12 was a later addition; the excavator He Nu argued for a 20-term seasonal calendar; Li Yong (2010) simulated sun-tangency sightlines over 3000–1000 BC to derive possible observing years; and Huang Kejia (2025) challenged the observatory interpretation itself. The open core — what the non-solstice backsights were for, what calendrical scheme (if any) the layout encodes, whether the structure was built or modified in more than one phase, and when it was built and used — remains unresolved and actively contested.

Background

The Taosi site in Xiangfen County, Shanxi Province, is a large walled settlement of the late Neolithic Longshan-period Taosi culture (c. 2300–1900 BC), often linked by Chinese scholars to the semi-legendary emperor Yao and his traditional capital Pingyang, and to the astronomical content of the canonical chapter 'Yao dian' (Canon of Yao) in the Book of Documents. In 2003–2005, excavators led by He Nu uncovered within the middle-period ceremonial precinct a large semi-circular stamped-earth (rammed-earth, i.e. earth compacted in layers within forms) platform, designated IIFJT1: a curved foundation of thirteen earthen piers, standing originally some 2–3 m above their surviving 4–17 cm foundations and separated by twelve narrow slots roughly 20–30 cm wide, arranged on an arc of about 11 m radius around a central observation point that excavation later confirmed to within 4 cm. Because the arc of the slots spans the range of sunrise azimuths at Taosi's latitude, and the distant Ta'er Mountain ridge to the east-southeast forms a natural foresight (the horizon feature actually sighted through the slots), the excavators proposed that the structure was a 'guanxiangtai' — an observation platform used to fix the seasons by watching where the sun rose through the slots. In archaeoastronomy a backsight is such a sighting slot or standing position used to view a rising or setting point on the horizon; the astronomical quantity fixed by a slot's direction is the solar declination (angular distance of the sun north or south of the celestial equator), whose extreme values are reached at the winter and summer solstices. A key physical circumstance is that the obliquity of the ecliptic — the tilt of Earth's rotation axis that sets the solstitial declinations, about 23.4° today — decreases slowly over millennia (about 47 arcseconds per century, the standard secular rate), so solstitial sunrise azimuths drift measurably over the four millennia since Taosi; this both enables and limits any attempt to date the structure from its alignments. Precise measurement of the surviving geometry (e.g. centerline azimuth 125.046° and vertical angle 5.809° for slot E2; 60.349° and 1.267° for E12) combined with this secular drift allowed Wu, Chen and Liu (2009) to show that at the archaeologically established date of the Taosi Middle Period (about 2100 BC, from radiocarbon evidence) the half-risen sun would have appeared inside slot E12, just to the right of its centerline, at summer solstice and exactly on the E2 centerline at winter solstice — an alignment better than at the present epoch, and compelling evidence that IIFJT1 was indeed used to observe solstice sunrises. Their analysis also exposed what remains unresolved. First, only two of the twelve slots are explained: the other ten (E1 and E3–E11) have no established function. Their centerline azimuths are approximately evenly spaced (about 5–6° in azimuth between adjacent slot midlines), dividing the sunrise arc between the two solstice slots into roughly equal spatial parts; Liu Ciyuan's companion Chinese analysis (2009) describes slots E2–E12 as roughly evenly dividing the annual sunrise directions and thereby producing a strongly non-uniform time system, with E1 related instead to the moon's southernmost rising extreme. Whether this geometry reflects any actual calendrical division of the year — for example a ten-month calendar of the kind attested for the later Yi people, or a scheme of roughly twenty seasonal terms — is a contested interpretive question in the literature rather than a conclusion of the published analyses: Pankenier, Liu and De Meis (2008) argued from the strongly non-uniform implied date intervals that the slots cannot demarcate equal calendar periods. E1 lies several degrees south of winter-solstice sunrise and can never have shown the sun, and E11 (together with E12) is visibly offset from the curve of the main foundation, raising the questions whether it had a different purpose and whether the structure was built or modified over more than one period. Second, dating: the solstice alignments fit the radiocarbon date of about 2100 BC only to a few arcminutes, and because centimeter-scale positional uncertainties at the roughly 11 m slot distance translate into century-scale date uncertainty under the slow obliquity drift, the astronomical date carries uncertainties of many centuries and cannot by itself fix the construction epoch against the archaeological date. Subsequent scholarship frames these questions within a broader comparative context (Pankenier et al. note the resemblance in purpose to the thirteen towers of Chankillo in Peru, built some two thousand years later) and within the later Chinese tradition of the 24 solar terms — the annual cycle of climatic marker dates that traditional East Asian calendars still use — whose origins a Taosi seasonal calendar would push back by more than a millennium.

Problem Statement

Determine the actual function and date of the rammed-earth structure IIFJT1 at the Taosi site — the overarching question left open by Wu, Chen and Liu's (2009) solstice analysis. Concretely, an answer must resolve, from the published survey geometry, the excavation record, and any new primary evidence: (1) the function of the ten non-solstice backsights E1 and E3–E11 — in particular whether their approximately even angular spacing, which divides the sunrise arc between the solstice slots E2 and E12 into roughly equal spatial parts, was intended to mark specific observation dates and thereby encode a real calendrical division of the year (for example a ten-month calendar or a scheme of roughly twenty seasonal observation dates, as competing interpretations propose), or whether the spacing is merely construction geometry with no calendrical intent — the implied observation dates being strongly non-uniform under either reading, or whether the slots served no astronomical purpose at all; (2) the purpose of the anomalous backsight E11 — which, together with E12, is offset from the curve of the main foundation — including whether E11–E12 were a later addition, i.e. whether the structure was built or modified over more than one period; and (3) the epoch of construction and of observational use, to the precision the evidence permits, given that alignment-based dating carries century-scale uncertainties and that published estimates of the observing epoch diverge from the radiocarbon-based archaeological date of about 2100 BC. A complete answer must adjudicate among the competing published interpretations — solstice-fixing ritual platform, 20-term horizon calendar, multi-phase construction, and the recent structural objection that the platform was not an observational instrument at all — and must remain consistent with the measured slot geometry, the horizon astronomy at the site, and the stratigraphic and contextual archaeological record.

References

  1. Wu JiaBi, Chen MeiDong, Liu CiYuan (2009). Astronomical function and date of the Taosi observatory. Science in China Series G: Physics, Mechanics and Astronomy 52(1):151–158. DOI 10.1007/s11433-009-0017-1. https://link.springer.com/article/10.1007/s11433-009-0017-1
  2. David W. Pankenier, Ciyuan Y. Liu, Salvo De Meis (2008). The Xiangfen, Taosi site: a Chinese Neolithic 'observatory'? Archaeologia Baltica 10:141–148. https://e-journals.ku.lt/journal/AB/article/1329
  3. Li Yong 李勇 (2010). 世界最早的天文观象台——陶寺观象台及其可能的观测年代 [The earliest astronomical observatory in the world: the Taosi observatory and its possible observing years]. 自然科学史研究 (Studies in the History of Natural Sciences) 29(3):259–270. https://www.cnki.com.cn/Article/CJFDTotal-ZRKY201003000.htm
  4. He Nu (2018). Taosi: An archaeological example of urbanization as a political center in prehistoric China. Archaeological Research in Asia 14:20–32. DOI 10.1016/j.ara.2017.03.001. https://doi.org/10.1016/j.ara.2017.03.001
  5. Huang Kejia (2025). A discussion of the astronomical observation relic at the Taosi site: Querying the present interpretation of the rammed earth structures excavated in IIFJT1 of the Taosi site excavation, Shanxi Province, China. Archaeological Research in Asia 42:100617. DOI 10.1016/j.ara.2025.100617. https://doi.org/10.1016/j.ara.2025.100617
  6. Sun Xiaochun (2014). Taosi Observatory. In: C. L. N. Ruggles (ed.), Handbook of Archaeoastronomy and Ethnoastronomy. Springer, New York, pp. 2105–2110. DOI 10.1007/978-1-4614-6141-8_215. https://doi.org/10.1007/978-1-4614-6141-8_215
  7. David W. Pankenier (2013). Astronomy begins at Taosi. Chapter 1 of: Astrology and Cosmology in Early China. Cambridge University Press, pp. 17–37. DOI 10.1017/cbo9781139017466.005. https://doi.org/10.1017/cbo9781139017466.005
  8. He Nu (2013). The Longshan Period Site of Taosi in Southern Shanxi Province. In: A. P. Underhill (ed.), A Companion to Chinese Archaeology. Wiley-Blackwell, Chichester, pp. 255–277. DOI 10.1002/9781118325698.ch13. https://doi.org/10.1002/9781118325698.ch13
  9. He Nu (2020). Longshan Culture Issues: Taosi and Cosmology. In: E. L. Shaughnessy, Paula L. W. Sabloff, Gideon Shelach-Lavi (eds.), The Oxford Handbook of Early China. Oxford University Press, pp. 138–157. DOI 10.1093/oxfordhb/9780199328369.013.7. https://doi.org/10.1093/oxfordhb/9780199328369.013.7
  10. Liu Ciyuan 刘次沅 (2009). 陶寺观象台遗址的天文学分析 [Astronomical analysis of the Taosi observatory site]. 天文学报 (Acta Astronomica Sinica) 50(1):107–116. NASA ADS bibcode 2009AcASn..50..107L. https://ui.adsabs.harvard.edu/abs/2009AcASn..50..107L/abstract (also Semantic Scholar CorpusID 130557510, https://www.semanticscholar.org/paper/2a7b187fb688686916d496299d584990f7e88a0f)