¹ Liangchao gangmu bei yao 兩朝綱目備要 (Taipei: Wenhai, 1967), 5.25a. Cangzhou qiaosou 滄州樵叟, Qingyuan dangjin 慶元黨禁 (Beijing: Zhonghua shuju, 1985), 1.27a. Dates are converted using the web-based tool provided by the Academia Sinica: http://sinocal.sinica.edu.tw/.

² Some sources (such as Liu's biography in Song shi and even Liu's tomb epitaph by Zhen Dexiu) mistakenly identify Liu's account as for a local school in Fuzhou 涪州 (Chongqing). That Liu's account was for Fucheng county in Tongchuan 潼川 (Sichuan) can be ascertained from the tomb epitaph for Yang Linggui. See Tuo, Tuo 脫脫 ed., Song shi 宋史 (Beijing: Zhonghua shuju, 1977), 397.12100Google Scholar; Zhen Dexiu 真德秀, “Liu gexue muzhiming” 劉閣學墓志銘, Xishan xiansheng Zhen wenzhong gong wenji 西山先生真文忠公文集 (Sibu congkan edition), 43.14b; Wei Liaoweng 魏了翁, “Hanzhou tongpan Yang jun Linggui muzhiming”漢州通判楊君令圭墓誌銘, Chongjiao Heshan xiansheng daquan wenji 重校鶴山先生大全文集 (Sibu congkan edition), 84.12a.

³ Jingnan, Shu 束景南, Zhu Xi nianpu changbian 朱熹年譜長編 (Shanghai: Huadong shifan daxue, 2001), vol. 2, 1353Google Scholar.

⁴ Lee, Thomas H.C., Government Education and Examinations in Sung China (New York: St. Martin's Press, 1985)Google Scholar. Han Fengshan 韓鳳山, “Tang Song guanxue zhidu yanjiu” 唐宋官學制度研究 (PhD diss., Northeast Normal University, Shenyang, China, 2003).

⁵ Yuwen, Zhou 周愚文, Song dai de zhou xian xue 宋代的州縣學 (Taipei: Guoli bianyi guan, 1996)Google Scholar. Zhifu, Tian 田志馥, Song dai Fujian miaoxue de lishi dilixue fenxi 宋代福建廟學的历史地理學分析 (Beijing: Jingji guanli, 2016)Google Scholar.

⁶ For a few examples, see Lü Xufeng 呂旭峰, “Song dai Henan difang guanxue yanjiu” 宋代河南地方官學研究 (Master's thesis, Henan University, Kaifeng, China, 2008); Cui Lijun 崔麗君, “Song dai Jiangxi jiaoyu yanjiu” 宋代江西教育研究 (Master's thesis, Nanchang University, Nanchang, China, 2007); Yang Jie 楊杰, “Liang Song Jiangxi de guanxue, shuyuan yu keju” 兩宋江西的官學、書院與科舉 (Master's thesis, Jiangxi Normal University, Nanchang, China, 2008); Ren Chao 任超, “Song dai Chengdu fuxue shulun” 宋代成都府學述論 (Master's thesis, Sichuan Normal University, Chengdu, China, 2017).

⁷ He Li 何李, “Tang dai jitiwen yanjiu” 唐代記體文研究 (PhD diss., East China Normal University, Shanghai, China, 2010), 36. For a discussion of the history of the ji before the Song, see Qian Lei 錢蕾, “Bei Song jitiwen yanjiu” 北宋記體文研究 (Master's thesis, Nanking University, Nanjing, China, 2014), 5–28.

⁸ The two anthologies are Fang, Li's 李昉 Wenyuan yinghua 文苑英華 (Beijing: Zhonghua shuju, 1966)Google Scholar and Xuan, Yao's 姚鉉 Tang wencui 唐文粹 (Shanghai: Shanghai guji, 1994)Google Scholar. He, “Tang dai jitiwen yanjiu,” 39–40.

⁹ For remarks of earlier scholars, see Jing, Hao 郝經 (1223–1275), Hao shi xu Hou Han shu 郝氏續後漢書 (Taipei: Taiwan shangwu yinshuguan, 1983)Google Scholar, 66A.31b, and Nai, Yao 姚鼐 (1732–1815), Zhujia pingdian guwenci leizuan 古文辭類纂 (Beijing: Guojia tushuguan, 2012)Google Scholar, “xumu,” 13ab.

¹⁰ Shu, Lin, Chunjue zhai lunwen 春覺齋論文 (Beijing: Renmin wenxue, 1959), 70Google Scholar.

¹¹ Lin, Chunjue zhai lunwen, 70.

¹² In addition to accounts and steles, I have also included two epigraphs (ming 銘): Zhang Jun's 張浚 “Xin xuemen ming” 新學門銘 and Zhang Shi's “Nanjian zhou Youxi xian xue Chuanxin ge ming” 南劍州尤溪縣學傳心閣銘. Both ming are preceded by a preface of considerable length, making them somewhat similar to commemorative accounts and steles. Zaozhuang, Zeng 曾棗莊 and Lin, Liu 劉琳 eds., Quan Song wen 全宋文 (Shanghai: Shanghai cishu, 2006)Google Scholar, 188:4137.139, 255:5743.432.

¹³ The first Confucian shrine under imperial auspices appeared in 385 and was on the premises of the Imperial University, and by the turn of the sixth century Imperial Universities in both northern and southern dynasties had shrines dedicated to Confucius.

¹⁴ Gao Mingshi 高明士, Tang dai Dongya jiaoyu quan de xingcheng: Dongya shijie xingcheng shi de yi cemian 唐代東亞教育圈的形成——東亞世界形成史的一側面 (Taipei: Guoli bianyi guan), chapter 1. Jinxing, Huang 黃進興, “Jiekai Kongmiao jidian de fuma: jianlun qi zongjiaoxing” 解開孔廟祭典的符碼——兼論其宗教性, in Wenhua yu lishi de zhuisuo: Yu Yingshi jiaoshou bazhi shouqing lunwenji 文化與歷史的追索——余英時教授八秩壽慶論文集, edited by Tillman, Hoyt (Taipei: Lianjing chuban gongsi, 2009), 535–58Google Scholar.

¹⁵ Meihua, Wang 王美華, Lizhi xiayi yu Tang Song shehui bianqian 禮制下移與唐宋社會變遷 (Beijing: Zhongguo shehui kexue, 2015), 141–45Google Scholar. Yinong, Cheng 成一農, “Song, Liao, Jin, Yuan shiqi miaoxue zhidu de xingcheng yu puji” 宋、遼、金、元時期廟學制度的形成與普及, in 10–13 shiji Zhongguo wenhua de pengzhuang yu ronghe 10–13 世紀中國文化的碰撞與融合, edited by Xiqing, Zhang 張希清 (Shanghai: Shanghai renmin, 2006), 166–72Google Scholar.

¹⁶ Gengyao, Liang 梁庚堯, Songdai keju shehui 宋代科舉社會 (Taipei: Taida chuban zhongxin, 2015), 76–77Google Scholar.

¹⁷ Yu, Han, “Chuzhou Kongzi miao bei” 處州孔子廟碑, Quan Tang wen 全唐文 (Beijing: Zhonghua shuju, 1983)Google Scholar, 561.5678a.

¹⁸ Xuan, Xu, “Sizhou chongxiu Wenxuan wang miao ji” 泗州重修文宣王廟記 (985), Quan Song wen 全宋文 (Shanghai: Shanghai cishu, 2006)Google Scholar, 2:25.232.

¹⁹ Hu Dan 胡旦, “Ruxue ji” 儒學記 (988), Quan Songwen 4:62.8. Duan Quan 段全, “Xianyou xian jianxue ji” 仙遊縣建學記 (1000), Quan Song wen 9:195.410.

²⁰ In a few cases, this was complicated by the fact that some inscriptions survived only in local gazetteers and their titles seem to have been added or modified by gazetteer compilers in later dynasties. For example, the title for Yuan Xie's inscription for the county school of Changguo was clearly added or modified by the Yuan dynasty editors. Yuan Xie, “Changguo zhou ruxue ji” 昌國州儒學記 (1224), Quan Song wen 281:6377.252.

²¹ These include thirteen inscriptions that were dedicated to the “shrine-school” complex (miaoxue 廟學).

²² For recent studies of the Southern Song academies and their close link to the Neo-Confucian movement, see Wenyi, Chen 陳雯怡, You guanxue dao shuyuan: cong zhidu yu linian de hudong kan Songdai jiaoyu de yanbian 由官學到書院——從制度與理念的互動看宋代教育的演變 (Taipei: Lianjing chuban gongsi, 2004)Google Scholar and Walton, Linda, Academies and Society in Southern Sung China (Honolulu: University of Hawai‘i Press, 1999)CrossRefGoogle Scholar. For an overview of academies in Chinese history, see Guojun, Li 李國鈞 ed., Zhongguo shuyuan shi 中國書院史 (Changsha: Hunan jiaoyu, 1994)Google Scholar.

²³ The inscription for the entrance of Changzhou county school is listed in Quan Song wen twice, each under a slightly different title and a different author due to conflicts in the sources. Both the inscription and the authors are therefore double-counted in this corpus. Qian Shuiyou 潛說友, “Changzhou xian gaili xuemen ji” 長洲縣改立學門記, Quan Song wen 347:8026.273. Qian Shuiyou 錢說友, “Xianchun gaili xuemen ji” 咸淳改立學門記, Quan Song wen 356:8260.427.

²⁴ These are “Fuzhou dudu fu xinxue beiming” 福州都督府新學碑銘 by Dugu Ji 獨孤及, “Kunshan xian xue ji” 崑山縣學記 by Liang Su 梁肅, and “Xiangzhou Kongzi miaoxue ji” 襄州孔子廟學記 by Pi Rixiu 皮日休. See Quan Tang wen, 390.3964a, 519.5275a, and 797.8354b. Note that one of them was dedicated to the “Confucian shrine-school” complex (Kongzi miao xue 孔子廟學).

²⁵ The vast majority of these inscriptions (725 out of 773) can be dated with a very high degree of accuracy on the basis of internal evidence. Dates for the remaining forty-eight inscriptions are estimated by adding fifty to the birth year of their authors. Fifty is the estimated average age at which the inscriptions in our sample were composed, and this is calculated from a total of 393 inscriptions which has been accurately dated and whose author's birth year is also known. The calculations yield an average age of fifty, with a median of fifty and a mode of forty-six. The author's birth year is known in twenty-five of the forty-eight inscriptions. For the remaining twenty-three inscriptions, the author's birth year is estimated based on where their writings appear in Quan Song wen, since the Quan Song wen compilers have arranged the titles by author and the authors by the best estimates of their birth years.

²⁶ Nevertheless, drawing on statistics in local gazetteers, a recent case study of Fujian shows that while the Renzong reign (1023–1063) saw a high tide in building local schools, very few schools were established between 1064 and 1127. See Tian, Song dai Fujian miaoxue de lishi dilixue fenxi, chapter 2.

²⁷ In the fourth month of 1094, the court changed the reign name from Yuanyou to Shaosheng, indicating the court's desire to reinitiate the reforms. Of the four inscriptions from 1094, three were composed after this change of policy and reign name. Thus, only one of the 1094 inscriptions is considered as written in the Yuanyou reign. However, one should keep in mind that the activities recorded in the other three inscriptions, in fact, were also carried out during the Yuanyou period.

²⁸ Tang Wenruo 唐文若, “Anyue xian xiuxue ji” 安岳縣修學記 (1143), Quan Song wen 199:4395.49. Chai Fu 柴紱, “Luling xian xiu xuegong ji” 廬陵縣修學宮記 (1144), Quan Song wen 198:4380.197. Yin Gong 尹躬, “Chongxiu Yongxin xian ruxue ji” 重修永新縣儒學記 (1145), Quan Song wen 179:3922.149. Sun Di 孫覿, “Lin'an fu Lin'an xianxue ji” 臨安府臨安縣學記 (1146), Quan Song wen 160:3480.375.

²⁹ Skinner, G. William, “The Structure of Chinese History,” The Journal of Asian Studies 44.2 (1985), 280Google Scholar.

³⁰ Many of these are also the same areas with a flourishing literati culture in the Song, which has been forcefully demonstrated in earlier studies of the geography of examination success. Chaffee, John, The Thorny Gates of Learning in Sung China: A Social History of Examinations (New York: State University of New York, 1995), 119–56Google Scholar.

³¹ Won-Hoi, Anthony Pak, “Towards an Analytical Approach to the Landscape Essay: Textual Analysis of Liu Zongyuan's Eight Records on Yongzhou,” Crossing Between Tradition and Modernity: Essays in Commemoriation of Milena Dolezalová-Velingerová (1932–2012), edited by Denton, Kirk A. (Prague: Karolinum Press, 2017), 61–85Google Scholar.

³² Qian, “Bei Song jitiwen yanjiu,” 82. He, “Tang dai jitiwen yanjiu,” 96–106.

³³ Qian, “Bei Song jitiwen yanjiu,” 82.

³⁴ Chen Shidao 陳師道, Houshan jushi shihua 後山居士詩話 (Baichuan xuehai edition), 7a.

³⁵ Some authors’ collected works, especially those that have survived in part or been reconstituted after they were already lost, do not include inscriptions for local government schools. For example, eighteenth-century editors of Siku quanshu reconstructed the collected works of Li Shi 李石 (b. 1108) after they were lost, but the reconstructed edition does not include any school inscription by Li. Both Li's school inscriptions studied in this article are preserved in a twelfth-century national anthology. Given the analytical purpose of this article and for the convenience of expression, collected works are considered non-extant if they do not contain any school inscriptions.

³⁶ Sometimes a prefecture and its subordinate counties fall inside different PMRs. In this study, PMRs are assigned according to the geographic coordinates of prefectural seats, so that men writing for a prefectural school and schools in its subordinate counties are not considered to be writing for schools in different PMRs. In any event, this affects only ten inscriptions in the corpus, five of which relate to the fact that the Huizhou 徽州 prefectural seat falls inside Lower Yangzi macroregion, while the seat of its subordinate county Wuyuan 婺源 is technically inside the Middle Yangzi macroregion. In addition, since most of North China was lost to the Jurchens during the Southern Song, the few places that technically fall inside the North China macroregion are reassigned to the Lower and Middle Yangzi in the Southern Song dataset. This affects seven inscriptions, including six for schools in Lianghuai (assigned to the Lower Yangzi macroregion) and one for a school in Jinghu (assigned to the Middle Yangzi).

³⁷ There are only five exceptions: Li Chui 李垂 (965–1033), Zhang Boyu 張伯玉 (fl. 1050s), Li Zhi 李廌 (1059–1109), Han Yuanji 韓元吉 (1118–1187), and Lin Yingyan 林應炎 (jinshi of 1235). In addition, Yu Hong's 余閎 home prefecture is unknown.

³⁸ It should be noted that authors who have inscriptions for schools in two or more PMRs are more likely to have a wenji that survives to today, but the nature of this correlation is ambiguous. On the one hand, the school inscriptions from men whose collected works have not survived are more likely to be a fraction of all they had written and these inscriptions are typically preserved in local gazetteers and local anthologies. The smaller number of their extant inscriptions and the geographical bias in the condition of their preservation may lead us to underestimate these authors’ scope of influence. On the other hand, we may reasonably assume that men of national renown are more likely to have a wenji and that their wenji are more likely to have survived to this day. If so, it is just as likely that not having a wenji that survives today was actually a product of the author's truly lesser influence.

³⁹ By chance, none of the writers in the corpus have extant school inscriptions both before and after 1126. Therefore, there is no overlap of authors between the two sets of inscriptions generated from the corpus.

⁴⁰ In this study, regional influencers in the same PMR by definition have direct ties to one another but no direct tie to regional influencers in any other PMR and, for that reason, always have a betweenness of zero. On the other hand, national influencers are always information brokers between different PMRs and have a positive betweenness. Therefore, what merits attention in this study is the betweenness value of each national influencer, for it indicates how important he was in the diffusion of ideas between different macroregions.

⁴¹ In this study these factors are measured by the number of other influencers in each macroregion for which the author functioned as a bridge, the number of extant inscriptions he wrote for schools in the macroregion, and the number of other national influencers who also functioned as a bridge for the macroregion.

⁴² 曾棗莊, Zeng Zaozhuang, “Keyou sanshi nian, bu chu Bo yu Ba: Chao Gongsu ji qi Songshan ji” 「客遊三十年，不出僰與巴」——晁公溯及其《嵩山集》, Tianfu xinlun 天府新論 6 (1989), 75–80Google Scholar.

⁴³ For a study of Li, see Xie Shanyuan 謝善元, The Life and Thought of Li Kou, 1009–1059. San Francisco: Chinese Materials Center, 1979.

⁴⁴ Song shi 421.12591.

⁴⁵ Hartman, Charles, “The Making of a Villain: Ch'in Kuei and Tao-hsüeh,” Harvard Journal of Asiatic Studies 58.1 (1998), 59–146CrossRefGoogle Scholar.

⁴⁶ Li Man 李曼, “Yijian Kongzi miao ji” 移建孔子廟記 (1094), Quan Songwen 80:1759.374.

⁴⁷ For some examples, see Quan Song wen 296:6741.66 [韶州州學師道堂記], 304:6951.318 [溧水縣建小學記], 288:6557.391 [鄂州州學四賢堂記], and 307:7018.361 [吳縣學慈湖先生祠堂記].

⁴⁸ For some examples, see Quan Song wen 275:6239.414 [二陸先生祠記], 288:6556.388 [徽州朱文公祠堂記], 294:6702.265 [鄉先生祠堂記], 304:6951.315 [南陵修儒學記], 307:7018.361 [吳縣學慈湖先生祠堂記], 319:7323.177 [臨川縣學勉齋祠記], and 319:7334.370 [三陸先生祠堂記].

⁴⁹ Zhu Xi, “Jiankang fuxue Mingdao xiansheng ci ji” 建康府學明道先生祠記, Quan Song wen 252:5653.61–62.

⁵⁰ Text analysis in this section is conducted in Exploratory Desktop, a data analysis program developed by Exploratory, Inc. See https://exploratory.io/.

⁵¹ For the most recent and promising development on word segmentation in Chinese texts, see Peng-Hsuan Li, Tsu-Jui Fu, and Wei-Yun Ma, “Remedying BiLSTM-CNN Deficiency in Modeling Cross-Context for NER,” arXiv:1908.11046 [cs.CL].

⁵² For two examples, see Paul Vierthaler, “Fiction and History: Polarity and Stylistic Gradience in Late Imperial Chinese Literature,” Cultural Analytics, May 23, 2016, and Sturgeon, Donald, “Digital Approaches to Text Reuse in the Early Chinese Corpus,” Journal of Chinese Literature and Culture 5.2 (2018), 186–213CrossRefGoogle Scholar.

⁵³ Of course, the two-character compound dali may also be the name of a state, but the nature of our corpus ensures that the term is rarely—if at all—used in this sense.

⁵⁴ The stopwords used in this study include 之, 以, 其, 而, 不, 為, 有, 也, 所, 然, 於, 則, 無, 矣, 曰, 此, 與, 焉, 未, 又, 乎, 于, 亦, 乃, 因, 且, 夫, 何, 盍, 哉, 耳, 豈, 不, 及, 若, and 如. Two-character terms beginning with zhe 者 are also discarded, but those ending with it are not. The interjection wuhu 嗚呼 is also discarded.

⁵⁵ For this purpose, Chinese calendrical terms are filtered, which include: all Song-dynasty reign titles; the sixty ganzhi terms; any two-character combination that contains the character 年, 月, or 日; and the terms 歲在 and 歲次 which often precede the ganzhi expression of a year. The only exception is Mingdao 明道, since it is only briefly used as a reign title and often appear in texts as the style name of Cheng Hao 程顥 (1032–1085). Also discarded are two-character combinations containing any of the characters that frequently appear in the corpus and indicate specific administrative levels: namely, 縣, 邑, 府, 州, 軍, 監, 郡, 令, 尹, 宰, 守, and 牧. This is to reduce the probability of documents getting grouped together simply because they all concern prefectural or county-level schools. Finally, combinations where both characters are Chinese numerals (一 to 十, 百, 千, and 萬) are also pruned from the list.

⁵⁶ A two-character combination that appears only once in the corpus is more likely to be a combination of two characters that are contiguous in a document only by chance, while a combination that has several hundred occurrences is unlikely to have formed only by chance. For example, sixian 祀咸 is a meaningless combination generated by the computer. It is extracted from the sentence desi xianyun 得祀咸允 (“[Their] reception of sacrifices are all appropriate”) and appears only once in the corpus. On the other hand, there are clearly meaningful combinations such as tianxia 天下, xiansheng 先生, xuezhe 學者, and xuexiao 學校, each of which has several hundred occurrences in the corpus.

⁵⁷ The algorithm of K-Means Clustering divides the documents into k number of clusters that minimizes the sum of squared distance between all vectorized documents within a cluster and the cluster center. The number of clusters (k) is fixed a priori. In this study, the value of k is set at 3 after experimentation with different k values and with reference to output from the Elbow method calculations.

⁵⁸ Of the sixteenth inscriptions from the 1020–1039 period, only two were composed in the 1020s, both in Cluster A. All the five inscriptions belonging to Cluster B were written between 1035 and 1039.

⁵⁹ Zu Wuze 祖無擇, “Caizhou xinjian xue ji” 蔡州新建學記 (1035), Quan Song wen 43:935.317.

⁶⁰ Wen Tong 文同, “Yongtai xian xinxiu shengmiao ji” 永泰縣新修聖廟記 (1070), Quan Song wen 51:1107.152. Lü Tao 呂陶, “Fuxue jingshi ge luocheng ji” 府學經史閣落成記 (1071), Quan Song wen 74:1610.50.

⁶¹ Li Bing 李邴, “Chuzhou jiaoshou ting timing ji” 楚州教授廳題名記 (1126), Quan Song wen 175:3823.64. Hu Cheng 胡珵, “Yanguan xianxue zhi ji” 鹽官縣學之記 (1135), Quan Song wen 182:3991.153. These inscriptions focus on two intellectual figures: Xu Ji 徐積 (1028–1103, disciple of Hu Yuan 胡瑗 [993–1059]) and Zhang Jiucheng.

⁶² Wei Liaoweng, “Chengdufu fuxue san xiansheng citang ji” 成都府府學三先生祠堂記 (1207), Quan Song wen 310:7094.259.

⁶³ Kongzi is among the top twenty frequent terms of Cluster C inscriptions, whereas tianzi is not even in its list of top fifty frequent terms.

⁶⁴ The term “wengong” 文公 on this list is ambiguous. As a posthumous title, it may refer to Zhu Xi but also Wang Anshi. However, it appears that in most cases, it is a reference to Han Yu (Han wengong 韓文公).

⁶⁵ The six inscriptions by Wang Anshi, one by Lü Huiqing, and another by Zhang Dun 章惇 (1035–1105) are the exceptions.

⁶⁶ These were Zhang Yu 張俞 (fl. 1040s), Hu Yin, Xie E 謝諤 (1121–1194), Hong Mai, Lou Yue 樓鑰 (1137–1213), and Cheng Bi 程珌 (1164–1242). All the inscriptions by these men fall in Cluster A, indicating that they wrote mainly on the general themes of Confucius, ritual offerings, and schools, without engaging intensively in the topics of the mid-eleventh-century statesmen or those of the Neo-Confucians.

⁶⁷ These three were Zhu Xi, Zhang Shi, and Wei Liaoweng. The other two were Ye Shi and Zhou Bida.

⁶⁸ Zhu Xi, “Longxing fuxue Lianxi xiangsheng ciji” 隆興府學濂溪先生祠記 (1179), Quan Song wen 252:5654.80; “Qiongzhou xueji” 瓊州學記 (1182), Quan Song wen 252:5655.91.

⁶⁹ Chen, You guanxue dao shuyuan, 27–106, 379, 389–95.

⁷⁰ When possible, the Neo-Confucians not only attempted to transform the ritual space of local government schools but also tried to transform their educational program in accordance with the Neo-Confucian vision. Chen Wenyi has provided a few cases from the thirteenth century in which Neo-Confucian scholars adopted the curriculum and pedagogy of the White Deer Grotto Academy after being appointed instructor of local government schools. You guanxue dao shuyuan, 186–88.