From the accidental discovery of a bronze “three-legged cauldron” (ding 鼎) during the reign of Emperor Wu of Han 武漢帝 in 116 BC, heralded as an auspicious dynastic omen, to the beginnings of epigraphic scholarship in the Song 宋 dynasty (960-1279 AD), ancient Chinese bronzewares have held a distinguished place in the Early Chinese material record. And it’s easy to see why. Many so-called “bronzes” – ritual vessels, bells, weapons, and implements – are thickly set with intricate, tortuously spiraling designs. We can only imagine the impression they must have made on their contemporaries when the copper in the bronze alloy had not yet oxidized, and these greenish wares were still golden and polished.
The earliest inscribed Chinese bronzes date from the Late Shang 商 dynasty (ca. 1200-1050 BC). These short (one or two characters) inscriptions typically featured clan insignia and the names of ancestral spirits. Following the Zhou 周 conquest of the Shang and their eastward colonial expansion, however, the scope and length of inscriptions expanded dramatically (the longest known inscription comprising some 500 characters). While early Western Zhou 西周 (1046-771 BC) inscriptions typically commemorated the investiture of the donor and the achievements of their ancestors, over time, bronzes came to be inscribed for quite disparate purposes – recording military achievements, marriages, family histories, land exchanges, and even musical theory.
With tens of thousands of inscriptions available to us today, digital databases offer a workable means of browsing, organizing, and interpreting the mass of information. In this series, we’ll explore Academia Sinica’s 中央研究院 bronze research tools, starting with the “Shang and Zhou Bronze Inscriptions and Bronzewares Database” (殷周金文暨青銅器資料庫). In part two, we’ll turn to look at the “Shang Zhou Bronzeware Geographical Information System” (殷周青銅器地理資訊系統), the “Lexicon of Pre-Qin Oracle, Bronze Inscriptions and Bamboo scripts” (先秦甲骨金文簡牘詞彙資料庫), and finally the “Bronze Inscription Related Document Database” (金文關係文獻資料庫). Some of the most valuable features in these databases are easy to miss, so I’ll be pointing those out along the way.
Academia Sinica’s “Digital Archives of Bronzewares and Inscriptions”
Sinica’s open-access “Digital Archive of Bronze Images and Inscriptions” (殷周金文暨青銅器資料庫) compiles some of the best resources currently available (see here for registration instructions):
The first database listed here – the “Shang and Zhou Bronze Inscriptions and Bronzewares Database” (殷周金文暨青銅器資料庫) – digitizes the 14,000 or so bronzes from the Late Shang dynasty to the Qin 秦 unification (221 BC) published in the Chinese Academy of Social Science’s two major collections, the “Yin Zhou jinwen jicheng” (殷周金文集成) or jicheng, and the supplementary “Xin shou Yin Zhou qingtongqi mingwen ji qi ying huibian” (新收殷周青銅器銘文暨器影彙編). These sources, incidentally, not only reproduce extant inscriptions, they also contain inscriptions and drawings of bronzes (many of which are now lost) documented by Song dynasty and later epigraphers.
Overall, the database has two major advantages. The first of these, the “Missing Graph System” (quezi xitong 缺字系統), we will discuss here while the second, the inbuilt GIS toolkit, will be discussed in part two. In terms of drawbacks, I’m hesitant to single out anything specific. As the database introduction explains, interpretations and transcriptions are not necessarily based on the most recent research, but they do represent mainstream scholarly readings. If there is one minor issue worth clarifying, it’s probably the nature of the interpretation search engine – so let’s begin there.
Search Functions
Search queries can be formed by combining the jicheng collection number (qi hao 器號), bronzeware name (qi ming 器名), periodization (tongqi shidai 銅器時代), inscription length (mingwen zishu 銘文字數), excavation site (chutu didian 出土地點), excavation period (chutu niandai 出土年代), current collection (xianzang 現藏), interpretation (shiwen 釋文), and publications (zhulu 著錄) – standard criteria across bronze inscription databases. Because bronzes are conventionally named using the formula “[donor] + [vessel type]” (e.g. Shi Qiang pan 史墻盤 = “the pan-basin commissioned by Scribe Qiang”), the bronzeware name search box can also be used to find inscriptions on specific vessel types.
To search inscriptions, we input one or more graphs into the “interpretation” (shiwen 釋文) box. Note that this box will perform an exact text search of interpretation fields which introduces issues for multiple graph searches. Consider the interpretation provided for an Early Western Zhou vessel known as “Recorder Xuan’s Wine Pot” (Zuoce Xuan you 作冊睘卣; jicheng 5407):
The interpretation field contains both a transcription of the bronze inscription in “Regular Script” (kaishu 楷書) style Chinese characters (underlined in green), and clarifications identifying the word written by these graph[s] (underlined in pink). Because transcriptions and interpretations are thus intermixed, searching for an expression like qian suxi 虔夙夕, “reverently [working away] from dawn till dusk,” will not return every meaningful instance of this phrase. To understand why, we need to know something about the nature of early Chinese texts.
Prior to early imperial standardization efforts, homophonous words often interchanged their graphic representation, and one graph might write two or more different words. This means that the graphs used to write “reverently [working away] from dawn till dusk”, might not be the same graphs we used in our query (虔夙夕). Consider the phrase as it appears on the “Lai Bells” 逨鐘 (Early Western Zhou): “虔
While frustrating, this multiple-graph search issue is hardly a shortcoming of Sinica’s database. Due to the orthographic complexities of Early Chinese writing, database editors must make a choice about how best to present inscriptions. Sinica’s solution reflects pragmatic considerations – the provision of in-line interpretations allows the user to swiftly make sense of an inscription. While a “fuzzy search” might better accommodate multiple graph searches, it’s usually not difficult to obtain the desired results by judiciously selecting a single graph from a string. Because su 夙 almost only ever appears in the expression “虔夙夕,” a query for this single character will return every instance of our phrase.
Another extremely useful feature is tucked away at the bottom of the search screen – the “Missing Graph Query” (quezi chaxun 缺字查詢 ) link.
This takes us to Sinica’s “Missing Graph System” (quezi xitong 缺字系統), designed to assist with inputting rare and variant graphic forms which we could not otherwise type into the database search boxes.
To locate a graph, we can search for one or more of its components. If we wanted to search for the unusual variant form of su 夙 encountered above,
The second column from the right represents graphs using a system specially developed by Academia Sinica – the “character composition formula” (gouzi shi 構字式). These formulae paruse characters into components and structural symbols. The structural symbol 
Returning to the rightmost column of our search result table, we find a “copy” (fuzhi 複製) button. This allows us to copy and paste the “constructed character form” into the bronze inscription search engine. As it turns out, all non-standard graphs in the database are actually construction formulae overlayed with image files of the graphs they encode. The database programmers have embedded a Java Script which identifies these formulae on a page and sends them to the formula database which then returns an image file with a transparent background to the formula’s original location. A major advantage of this is that we do not need to download a special font to read the transcriptions, as is required for so many other Early Chinese text databases. The entire character composition system has also been made freely available for integration into other websites. One cannot help but wonder whether this kind of technology might be useful for other kinds of Early Chinese text databases – particularly given the difficulties with creating Early Chinese fonts.
Returning to the search engine, let’s suppose we want to find inscriptions on bronze bells that include the word yue 樂, “music, dance.” Entering zhong 鐘 “bell” into the bronzeware name field, and yue 樂 into the interpretation field, we get the following 107 results:
At the top are several options for displaying results chronologically or by jicheng collection number. Clicking a result opens a detailed viewing page:
The upper window informs us the bronze was excavated from a “hoard” (jiaozang 窖藏) at the Famen Temple site in Fufeng County, Shaanxi 陝西, and is currently housed at the Shaanxi Zhouyuan Museum (陝西周原博物館). Images are provided with a sketch of the artifact, giving us a better three-dimensional sense of the object. The lower window features a copy of the original rubbing and “inscription location” (mingwen weizhi 銘文位置) data, the interpretation itself, and related publications with page numbers helpfully provided.
Overall, credit must be given to the database’s “ecosystem” design. The integration of different research tools allows for fluid movement between geographical, archaeological, and interpretive information. Together these resources constitute a powerful arsenal in the Early China scholar’s pocket and will no doubt contribute significantly to the quality and scope of future digital research. No Early Chinese text database can skirt issues of inscription representation and interpretations in so far as these impact search functionality. Perhaps the best we can ask is that new interpretations might one day be included in Sinica’s system.
The Digital Orientalist 