Digital Anthropology and the Renewal of Waishengren Studies : From Digitized Tombs to Identity Claims

In this study we illustrate the potential contribution of digital archives to the study of variation among Taiwan’s Waishengren. Using the digital archive ThakBong of gravesites in Taiwan, into which the census data of 1956 have been merged, we show that Waishengren varied in their practices and their social structure from North to South and through time in yet unobserved patterns. Our interpretation is that Waishengren assimilated and potentially even merged into existing Holo communities in the South, while in the North, Waishengren developed social distinctions through these practices to reflect power and ethnicity. Regional distinctions will thus be central in future Waishengren-studies.


Introduction
Waishengren ("Mainlander") is the term used most frequently in Taiwan to refer to those Han Chinese migrants who arrived from Mainland China after the Republic of China (ROC) took over the administration of Taiwan.In English, the word "Mainlander" is ambiguous and might refer either to daluren, i.e. a Han Chinese living in Mainland China, or Waishengren, i.e.Taiwan's Mainlander.In this paper we thus consistently use the term Waishengren.

Waishengren in Taiwan
After 1945, the Japanese had surrendered and ROC had taken over the administration of Taiwan and Penghu (the Pescadores).The last migration wave from China to Taiwan gathered momentum.The majority of these Waishengren arrived in winter 1949 from different parts of China, from different ethnic groups, including Han Chinese, Hui and Manchu, and from different religious beliefs, such as Buddhism, Taoism, Christianity and Islam.Many different streams of refugees continued to arrive in Taiwan along very different paths until the 1980s.While Kuomintang (KMT) soldiers kidnapped young men and war prisoners of both the Chinese Civil War and the Korean War, that were forced to move to Taiwan, some people deliberately came to Taiwan for fear of communists.Among them, were Korean War land owners, bankers, intellectuals, religious minorities, masters of traditional arts and artists.Many of them settled in Taipei City, where KMT had installed the ROC government.
The arrival of the Kuomintang caused traumatic changes to the social and cultural life of the local Taiwanese.They had, by force, to adapt to a new administration, with Standard Modern Mandarin as the new official language, to the new Republican Calendar and a new economy.Even worse, native Taiwanese, who had acquired the Japanese language, had studied in Japan, had fought in the Japanese army and adopted Japanese practices, saw the Waishengren's hatred towards Japan turning against them, leading to the 228-incident and the subsequent period of White Terror.
For many lower-class Waishengren, the situation was not easy neither.Not only were they not spared from the terror with which the KMT purged the new territory.Separated from their homes, without contact to family members in China or Taiwan, wounded or traumatized through war, many Waishengren lived in hastily constructed shelters, sometimes without permission to leave the army or to marry.Not partially beloved by the native population, for decades they could not develop any promising life perspective and lived among themselves in army villages or veteran dormitories, if KMT cared for them.Others fell out of the army and lived in extreme poverty (Yang and Chang 2010).

Waishengren Studies
It was only in the early 1980s, that the Waishengren became a research topic in the social sciences, although they had been identified earlier by foreign scholars as a social group.Early researches focused mainly on issues of ethnic discrimination institutionalized by and for the benefit of the Waishengren.As the epistemological context remained at the turn of the 1990s intimately related to developments in politics and society, studies started to tackle questions of ethnicity, the identity crisis of Waishengren (Yang and Chang 2010) and sub-groups differences, for example, among Waishengren (Hu 1989).In the last decade, the study of subgroup differences and the more recent and global paradigms of diaspora and gender studies became the main driving forces in the study of Waishengren (e.g.Scott 2006).
Yang and Chang (2010), most recently, in their programmatic paper on Waishengren, identify the need for a paradigmatic shift in the study of Taiwan's communities to facilitate a comprehensive survey of them, including the study of inner-group variations.This paradigmatic shift in research, claim the authors, requires a re-evaluation of research methodologies, in the process of which archives would assume a central role.The authors, however, did not elaborate the relation between archives as main data sources and a new research focus on innergroup variations.

Towards Digital Archives
Our interpretation of this claim is that only archives and potentially only digital archives can fulfil the requirements of size, collaboration in data creation, and open access.Studies of inner-group variations require data sets, the size of which grows exponentially with the number of variables along which variations are examined.A study involving three types of social classes in three regions and through three time periods would require a twenty-seven times larger data set than an approach that would ignore regions and time periods.Second, the construction of such large data sets requires more manpower and skills than a single researcher can contribute in his or her lifetime.The required interdisciplinary skills for this endeavour range from information science, knowledge of informants' languages and writing systems, to the knowledge of relevant theories in social sciences.Third, this large data-set needs to be made accessible to people who are not involved in data collection, to put the archive into use beyond the life span of the involved researchers already while the researchers are still involved.Archives thus are more than dusty vaults filled with hand-scribbled scrolls.Archives are collaboratively constructed data-sets that are made accessible, in portals through meta-data, to a wider research community, mainly in the form of digital resources.
Through the digitizing of data, traditional studies in the humanities will be enabled to keep pace with the digital revolution of the society and its citizens (e.g.Moretti 2005;Genet and Zorzi 2011).Through digital data, the traditionally paperbased documents are not only rendered searchable for computers, but objects such as houses, temples, or tombs can now be turned virtual and ready for archiving.A research based on digital archives hence enjoys the advantages of intersubjectivity, speed and accuracy.In particular, the social scientific conflict between the general and the specific can now be freshly tackled.Human brains as exclusive research tools are sub-optimal: our phylogenetic endowment wants us to draw conclusions from one crocodile attack as opposed to a statistically significant number of crocodile attacks, and thus tends to over-generalize (Vollmer 1975).Powerful computational analyses, such as geographic analyses or time-line analyses are exempt from this tendency.They can be performed for hundreds of places, communities or time periods, checking the validity of any generalization.The creation, preparation and analysis of these digital data, we call, respecting the thematic field in which we work, "digital anthropology".

Thakbong, the Digital Archive on Taiwan's Gravesites
The digital archive we develop along these lines is an archive of Taiwan's gravesites, called ThakBong.The ThakBong Digital Archive Project aims at a representative digital documentation of Taiwan's gravesites in the form of their tombs and tombstones.Gravesites, tombs and tombstones are documented through digital, geo-referenced images, as well as transcriptions and descriptions of the photographed objects, formalized in XML (Extensible Markup Language).Tombs thus become searchable like web-sites through key words like a family name, the life data of a deceased, a placename written on the tombstone, symbols, forms, offerings and the featurized architecture of a tomb.Through the search of two or more features, correlations between features can be established, which represents the basic methodological approach in the analysis of such data, similar to styles of analysis in Corpus Linguistics (Bieber et al.1998).
Up to now, about 30.000 tombs have been documented through 100.000 photos.The full annotation of the tombs is an ongoing process and will take years to be completed.The sampling of the gravesites tries to match the proportions of Taiwan's census data of 1956, as represented in Chen and Fried (1967), in terms of the number of people per administrative region, and within a region, their socalled ethnic, but de facto linguistic distributions (Streiter et al. 2011).Besides gravesites in Taiwan, the ThakBong archive documents a limited number of gravesites outside Taiwan, such as the People's Republic of China and places where Chinese settlers have moved to, among them, Hong Kong, Hawaii, USA, the Philippines and Europe.

ThakBong at Work
In this study, we seek to respond to Yang and Chang (2010) and to show how archival studies can contribute to Waishengren-studies.More precisely, we analyze a few aspects of the epigraphical and funeral practices of Waishengren documented in the ThakBong digital archive, with a variationist approach, and correlate variations of tomb-features to space, time and social classes (Streiter et al., 2009(Streiter et al., , 2010a(Streiter et al., , 2010b)).In all data presented in this section, we compare mainly three groups, "Mainlander", "Holo" and "China".Here "Mainlander" (Waishengren) and "Holo" refer to the respectively non-Christian and non-Muslim Han ethnicities.The reason why we exclude Christian and Muslim tombs from our analysis is that tombs of these religions are heavily framed by features of religious ingroup similarity that outweigh the impact of ethnicity."China" refers to our small collection of tombs on the territory of what is now the Peoples' Republic of China.

Waishengren: Population Size, Gender Ratio and Family Names
One of the first problems for quantitative research of Waishengren is the question of how many Waishengren had migrated to Taiwan.For socio-political reasons, precise data have remained secret, but most estimations are about one million people or more.The 1956 census (Chen and Fried 1967) reports 914.000Waishengren in Taiwan and Penghu, making up 9,8% of the overall population.However, the census does not report data from Kinmen and Mazu.The ThakBong tombstone corpus describes the tombs of about 30.000 people for Taiwan, Penghu and Kinmen, of which 12% is unambiguously marked as Waishengren.The actual numbers might be higher, as the default assumption for this annotation feature is Holo and so the number of Holo tombs may include either tombs that have not yet been annotated or Waishengren tombs that have been assimilated to the Holo styles.Overall, however, the current sampling does not deviate from the most common estimates nor the 1956 census data.
The percentage of Waishengren in the overall population in 1956, according to Chen and Fried (1967), is shown in Table 1.This table lists locations with a high percentage of Waishengren first.The largest number of Waishengren lived in the main cities.Taipei and Keelung, of the cities, and Taipei County and Hualian County, of the counties, hosted most Waishengren.The smallest amount of Waishengren was in the south-western counties of Zhanghua, Yunlin, Chiayi and Tainan.These numbers might, to some extent, characterize the relations that Waishengren maintained with their social environment.As for epigraphical and funeral practices, this might mean that Waishengren were under a strong pressure to assimilate in places such as Yunlin.Their population size and their economic and political power might allow them to resist assimilation pressures further to the north, or allow them even to impose their own practices on other groups there.

Location
Percentage Another feature that allows to link the ThakBong data to other Waishengrenstudies is the gender ratio, defined as the size of the male population divided by the size of the female population.As expected, the literature reports a very unbalanced gender ratio for Waishengren, as most of them were young soldiers.Li Tung-ming (1970), for example, reports an unbalanced gender ratio, reaching almost five males for one female at the age of 40 years at the time of the census.Table 1 reports gender ratios as estimated from tombstones.High gender ratios seem to correspond to the soldier profession and reduced economic and political power.In our future research we will thus try to correlate, through tombstone data, the gender ratio to markers of professions and to the relative wealth, as expressed through the tomb and its location.The data from Kinmen, included in ThakBong, but not in the 1956 census, report a remarkable gender ratio of 117.
Figure 1, shows the comparison of the gender ratio of Taiwan's Han groups through time, as revealed through tombstone inscriptions.The gender ratio of Waishengren in Table 1 above was from 1950 and of Holo or Hakka communities and has remained so ever since.The difference between the gender ratio in the cited population statistics and the gender ratio in tombstone data is potentially due to the fact that many Taiwanese women might have been buried with the Jiguan (the geographic origin of Mainland China) of their husband, and thus have been classified as Waishengren.
Our tombstone data in addition reveal whether or not Waishengren had offspring, thus allowing to correlate offspring, place of settlement and the place of origin.The latter will be discussed below in detail.Family names can be finally linked to the place of origin, the place of settlement and the size of offspring.Our preliminary data on the relation between settlement and family names show that different names have very different types of patterns where they settled in Taiwan.While some Waishengren family names are literally randomly distributed over the island, for example Gu3 (古), as shown in Figure 2, other family names form regional clusters, for example Tu2 (涂), as shown in Figure 3. Plotted are the 5% of townships and city districts with the highest percentage of a family name within the group of Waishengren.In Figure 2-3, the right part indicates Ripley's K.The distribution is random when the solid black line is within the two red dotted lines.The distribution forms a statistically significant cluster, when the solid black line is above the red dotted lines (Crawley 2007, 754-758).

Waishengren: Heterogeneity of Practices
In daily life as in scientific literature, the term "Waishengren" is often used to refer uniformly to a group of people that actually have less in common than suggested by the denominator.De facto, Waishengren were originally defined only through their shared history of migration in the aftermath of Chinese Civil War.Despite this shared history, the context of the migration, the force that made them come to Taiwan and, as a consequence, the subjective experience were very different, and so were their age, their family status, their economic status and their role during the White Terror, as victim or perpetrator.Nonetheless, as the term "Waishengren" continued to denote the offspring of the generation that had migrated, the definition and connotation of the term started to shift away from the history of migration and acquired administrative and ancestral meanings.The diversity of epigraphical and funeral practices of this group, as documented through tombs, questions any conception of a uniform or static social group.Figure 4-6 show a subset of the great variation of Waishengren tombs one can observe.This variation does not affect arbitrary or marginal tomb-features, but highly symbolic features, such as the color of the characters, the writing direction, the language, symbols, offerings, tomb style, the degree of assimilation to local tomb, the relation to the ground, and the positioning of the coffin.Photos by the authors.
We claim, that these tomb-features are crucial for the understanding of differences and similarities among Waishengren.Tombs are related to practices: Tombs are created by practices and are made to serve practices.And practices are important for an analysis of differences or commonalities in Han Chinese societies.Practices have been used throughout the Han Chinese history to perform cultural assimilation and assume loyalty through the principle of eupraxia, the conform public behavior.Beliefs or rational explanations behind practices are peripheral to the definition of ingroups and outgroups in Han Chinese societies (Watson 1988).Through the displacement of Waishengren, however, different linguistic, religious and cultural practices were brought together in one group, which as a minority had to find its place in the new society.As a consequence, a transformation and potential standardization of the diversified practices became of vital interest for this group to meet their different needs, such as ingroup (Waishengren) and outgroup (local people) support.The question whether and how such a standardization of diversified practices developed lies naturally at the heart of Waishengren-studies as this provides a developmental perspective to the facevalues obtained, for example, in questionnaires.

The Limit of Uniformity: Power
While the standardization of language and folkloric culture was promoted in ROC through education, other diversified practices, such as epigraphical and funeral practices, were difficult to manage.Although KMT achieved sometimes a high degree of uniformity within a graveyard it had the power to manage, variation among graveyards persisted, as seen in Figure 2. Despite the fact that no standardized common tomb-style for Waishengren developed, individual tombstone-features developed within different directions to meet different needs, among them, the need to express a Waishengren ingroup identity.The standardization of tomb-style for Waishengren was impossible to achieve as long as different sub-groups, the rich and poor, the assimilated and the elites had different requirements.
One tomb-feature that had the potential to develop into an ingroup feature for all Waishengren was the form of the tombstone.The two most common tombstone forms, as shown in Figure 4 to 6, are the rectangular tombstone and the toprounded tombstone.As the top-rounded tombstone was mainly used by local people, Waishengren could have assumed the freely available rectangular tombstone form as their ingroup feature.Indeed, we observe in Figure 4c), Figure 5a) and Figure 6a) and b), that the rectangular form developed for some Waishengren into a programmatic feature, repeated in different aspects of the tomb, such as a rectangular ground plan or a rectangular Jinlu (ghost money stove).
Figure 7 provides a more detailed analysis of the development of this feature.Figure 7a) sketches the temporal development of the relative usage of the rectangular versus top-rounded tombstone for the categories "Mainlanders", "Holo" and "China".The Holo prefer the top-rounded tombstone.Only during the Japanese period, during which some tombstones resembled, in form and material, bricks, rectangular tombstones have been used to some extent.For the category "Mainlander", we observe a much more consistent use of rectangular tombstones.In our limited data-set "China", almost all tombstones are rectangular."Holo" and "Mainlander" show sub-group differences when divided into "North", "Middle" and "South".In the North, the Waishengren chose the rectangular tombstone only after 1970 and the Holo followed this trend.In the South, the Waishengren chose from the beginning the local form.In the middle, Waishengren gradually replaced the rectangular tombstone by the top-rounded tombstone.Figure 7b) shows the regional distribution of the rectangular tombstone, compared to the top-rounded tombstone for the two groups Waishengren and Holo in different counties.These data confirm that there are important regional differences in the use of rectangular tombstones.With the exception of Taichung, where the form of Waishengren and Holo tombstones oppose each other, Waishengren and Holo achieve a locally negotiated consensus on the tombstone form.In Taipei City, Keelung and Hualian, where Waishengren were most numerous, Holo assimilated the rectangular tombstone form.In the South-West, Waishengren adopted the form of the Holo.While the distribution of the rectangular tombstone can be empirically approached, its meaning is a question of interpretation.It is certain, that this form, common in rural China, opposes local practices in Taiwan, where the rounded top is believed to improve the feng shui of the tomb.This however does not warrant the interpretation of this tombstone-feature as a feature traded from China: Where the feature is used most commonly, it has been re-invented not earlier than 1970.A second, alternative interpretation, as to which the rectangular tombstone represents a rational world view, which does away with notions of superstition, such as feng shui, can be dismissed also, as many Waishengren tombs are placed and properly oriented in expensive ground, the essence of which is its excellent feng shui.We thus attempt an interpretation of this tombstone-feature in relation to the social situation in Taiwan: We hypothesize, that the geographic patterns of this feature match the social relation between Waishengren and the local population: Waishengren in relative minority positions and in need of outgroup support assimilate local tomb styles.Higher class Waishengren, on the other hand, try to distinguish themselves from local people through a tomb style that does not reflect the concerns of small people of how to improve an otherwise mediocre feng shui.We thus understand the rectangular tombstone of upper class Waishengren and local people as a symbol and an ingroup feature of the rich and powerful, who had unrestricted access to restricted land, defying the policy of cremation imposed on small people, defying unreasonable prices and defying plans of land usage and landscape protection.Thus, this feature did not develop into an ethnic Waishengren ingroup feature, but into a feature representing power.The function of a Waishengren ingroup feature was to be taken over by another tombstonefeature, which even coopting local people would not adopt.

Epigraphical Practices of Waishengren: The Provincial Origin
If Waishengren are and have been perceived as a unified and homogeneous group, this is mainly due to social and discursive construction, thus, a result of the label "Waishengren" itself.Through its continuous use, from Chinese Civil War until today, the term "Waishengren" acquired and lost meaning components, through which the term and the conception of the designated group was permanently reshaped: During the Chinese Civil War, the term indicated a soldier with a provincial origin that is different from where he was stationed, for example a man from Guangdong in Shandong.In Taiwan, the connotation was later reduced to a bipolar reading, opposing Waishengren to the so-called Benshengren, locally born people.The term thus indicated a person originating from a province other than Taiwan.The term became institutionalized as a ethnic census category when applied also to the children of Waishengren, but not to their local wives.From this census categorization resulted a social and ethnic discriminatory segregation, reinforced by the self-representation of ROC that, albeit relocated to Taiwan, still claims the whole Chinese Mainland (Wang Fu-chang 2005).Access to universities, for example, was regulated through quota for all provinces, of which Taiwan was just one.The Jiguan-based categorization ended in 1992, when ROC institutions started to be indigenized and the preferential treatment of Waishengren was no longer timely.When the term lost its administrative value, it could fully assume the meaning of an ethnicity.This new ethnic classification in Taiwan puts Waishengren as one Han Chinese group beside the local Holo and Hakka (Wang 2008, 510).
The fundamental marker by which Waishengren claim a unique status and their identity was the belonging to a Jiguan, the geographic origin in Mainland China.The Jiguan was also a crucial factor in the social construction of this community and its labeling (Corcuff 2002, 170;Yang and Chang 2010, 112).According to our understanding, both notions, Jiguan and Waishengren, emphasize unity and difference at the same time in two different semantic fields.Unity is created through the superordinate notion of "guo" (country), which is implied by the notion "sheng" (province) within the term "Waishengren".Unity is also created through the notion of homeland in China, of which the Jiguan is an instantiation.Thus, while the term Waishengren assumed a bipolar meaning, the term Jiguan preserved the multivalued scheme.Formally, both notions are linked through the notion "sheng", as a category in "Waishengren", as a value in the Jiguan.
Similar to the institutionalization of the Jiguan in census, the term also became an almost standardized way to identify a Waishengren through tombstones.As the notion of Jiguan had already existed in Taiwan before the migration of Waishengren, the exclusive use of the Jiguan by Waishengren required a complex negotiation among all social groups: The offspring of Holo and Hakka that had migrated between the 17th and 19th century from China had always kept the memories of their origin in more or less fine-grained Jiguan-based distinctions, such as Quanzhou or Zhangzhou, inscribed on houses, family temples and tombstones.The common usage of the Jiguan by all Han communities, as shown in Figure 8 was thus not unproblematic, especially when the Jiguan developed into a marker for Waishengren.As a consequence, Holo and Hakka continued a shift they had initiated during the Japanese administration, when they replaced the Jiguan on the tombstone with other placename types, probably as a reaction to the Japanese classification of people into "Fujian" (Holo) and "Guangdong" (Hakka) in census and administration.These alternative placename types are local Taiwanese placenames and the Tanghao.Especially the Tanghao became popular after 1949, as this allowed Holo and Hakka to place a reference to China, but unlike the way the Waishengren did.Waishengren.Photos by the authors.
The Tanghao is an ambiguous notion, relating to place and family.The primary meaning of the Tanghao is a placename, mostly in Northern China, where the ancestors of a family are assumed to have originated from, 1000 to 2000 years ago.In other words, the Tanghao refers to the probably mythic homeland of the patronym bearing the ancestor who started to move south form the Central Plains in China.The secondary meaning is the distinction of family branches with the same patronym, through the different histories implied by the Tanghao.Taiwanese who use the Tanghao in this latter meaning are not necessarily aware of the spatial meaning of the Tanghao.
Figure 9 illustrates the fundamental change that took place within one century in the use of placenames by Taiwan's Holo communities, paving the way for the use of the Jiguan by the Waishengren.Meanwhile, Jiguan, if continued to be used by local people, shows up as a two-character denomination that does not mention the province, see Figure 9 a) and b).The Waishengren, on the other hand, adopted a four or six character style for writing their Jiguan, that systematically mentions the province through the first two or three characters, cf. Figure 9c).Through the provincial origin other than the provinces the local Taiwanese claim as their origin, i.e. mainly Fujian and Guangdong, and the implication of the superordinate notion of "China", they mark themselves as Waishengren.Waishengren from Fujian and Guangdong thus had some degrees of freedom to identify as Waishengren or to assimilate to the local community.Figure 10 shows the spatial distribution of Waishengren origins, according to the indications written on tombstones, sampled in ThakBong.The comparison of the graph in Figure 10 and the graph on Waishengren origin shown in Li (1970, 64) reveals a correlation between the size of circles derived from tombstone inscriptions and the size of circles in the demographic study conducted by Li.However, we observe that Waishengren originating from Fujian do not appear as frequent on the map derived from tombstone data.It is assumed that the tomb-based classification cannot distinguish them clearly from the Holo, whose Jiguan is also in Fujian.However, as the ethnic classification in ThakBong is not based on the reference of the Jiguan alone, but also on the forms of inscriptions (2-character Jiguan vs. 4 or 6-character Jiguan), the Waishengren from Fujian must have assimilated to the Holo tomb-features to some extent, while giving up those Waishengren features, which are indeed established Waishengren markers.
The problem for Guangdong is different, because many Taiwanese Hakka, although originally from Guangdong, do not necessarily place a Jiguan on their tombstones.Instead, they frequently mark the familial generation, starting from the migration to Taiwan, usually a number between 10 and 30.In the regions where the Hakka do mention a Jiguan on tombstones, the much more formulaic inscriptions and the form of the tomb of Hakka make it possible to easily distinguish them from those of Waishengren with the same Jiguan.

Funeral Practices of Waishengren: Body and Earth
Funeral practices are classical topics in anthropology and deserve particular attention in any cultural and social context, especially in the cases of migrants who were buried by their community in a foreign country (e.g.Benninghausen 2005).Quite problematic is the situation for Chinese migrants, if they had departed without the intention of settling down, because, according to traditional Chinese culture, the corpse has to be buried in one's homeland.Burial practices, and the specialized social organizations accompanying them, have thus been established in Chinese migrant communities from the end of the 19th century in America until today, which include a first burial, an exhumation, the transport back into the homeland, and a reburial (Chung and Wegars 2005).
Confirming most studies about ethnic communities in Taiwan, we observe distinct and changing practices of Waishengren compared to other communities: A substantial part of Waishengren tombs are designed for the first burial only with tombs that can be easily opened and coffins stored above the ground to prevent them from rotting (cf. Figure 11).In this burial mode, the coffin is placed above the ground and a structure is built around it.The size of this surrounding structure identifies a second feature of this burial mode: The coffin is not in direct contact with the building material or the earth.As a consequence, the coffin will remain easily manipulable for a long time.Second, this burial mode might lead, wished or not wished, to a mummification of the corpse instead of its decomposition.This mummified corpse evokes the legend of the walking mummy that finds its way back to the homeland, or daolushi (road-finding-zombi), told in some regions of China.
The interpretation of this burial mode must be seen in relation to the Chinese tradition to be buried in the ground of one's homeland.These tombs deny Taiwanese ground the status of a homeland and keep the coffin manipulable for possible relocation of the corpse.However, as conceptions and self-conceptions of Waishengren have changed, it is worth to explore how funeral practices have been adapted.For instance, while Chiang Kai-Shek, since his death in 1975, has not be buried and his coffin is waiting above the ground to go back to his homeland, Ma Ying-jiu repeatedly affirmed that he will be buried in Taiwan.Thus, how Waishengren negotiate this problem of ultimate settlement naturally turns into a crucial question for Waishengren-studies.Although the amount of data available at the moment is not sufficient to firmly establish such differences, our research on inner-community variations in Waishengren funeral practices tends to highlight the heterogeneity in this community.The alternatives to the burial above the ground are the burial, at which the coffin is covered by earth and stones.This is the common practice of Benshengren to facilitate the decomposition of the corpse and thus to prepare the second burial.Another alternative, practiced almost exclusively by Waishengren, is to store the coffin in an underground compartment, covered by removable slates.The effect on the corpse and the coffin in this burial mode might be similar to the storage above the ground, but the most visible difference with respect to the tombs of Benshengren has been overcome.Clustering 50 randomly chosen Waishengren tombs with features related to the position of the coffin (above, covered, inside), the form of the tombstone and the presence or absence of a placename, we obtain the tree-structure shown in Figure 12.This tree-structure suggests a principal two-or three-partition of epigraphical and funeral practices and makes it possible to associate each of these clusters with a conception of how the notion of "Waishengren" is defined, that is through a) migration from China after 1945, b) an administrative classification or c) an ethnic classification.To do so, we characterize these clusters through typical values or average values for each cluster (Romesburg 2004).Doing so we obtain the clusters: I (25 items): 1989, covered, top-round, North 24.026 degree, II (11 items): 1973, above the ground, North 24.741 and III (14 items): 1979, inside underground compartment, rectangualr tombstone, North 24.246.Thus, the development of funeral practices of Waishengren went from "above" to "inside" to "covered".Second, "covered" is more common in the South of Taiwan, similar to the toprounded tombstone.The intermediate funeral practice "inside" is located further north.The most conservative burial form "above" can be found mainly in the very North of Taiwan.The rectangular tombstone form and the burial mode above the ground do not perfectly correlate.Instead, the rectangular tombstone seems to correlate better to the administrative conception of Waishengren.Waishengren tombs in two or three classes.Below the tree-structure, the tomb identification and the date of tomb erection are listed in addition to the variables used for clustering.The labels "covered", "above" and "inside" describe the position of the coffin, "tr" means top-rounded tombstone, "4c" means rectangular tombstone and "loc" means placename in topic position.
If these three clusters and the three definitions of Waishengren can be really equated, we would equate (a) to II, (b) to III and (c) to I, and conclude that the idea of Waishengren as a third Han-ethnicity in Taiwan has been accepted firmly only by the Waishengren in the South of Taiwan, while those in the North might still retain the idea that they are Han who came 1945 from China and will return to their homeland.Further research however will be needed to confirm the differences in practices among the North and the South in Taiwan, and whether this corresponds to self-conceptions of Waishengren.Also of interest for Waishengren-studies is Taichung, where the notion of power outweighs and transcends that of ethnicity.

Conclusion and Perspectives
In this study we illustrate the potential contribution of digital archives to the study of variation among Taiwan's Waishengren and thus to answer to the appeal of Yang and Chang (2010) to renew Waishengren-studies.Using the digital archive ThakBong of gravesites in Taiwan, to which the census data of 1956 have been merged, we applied time-line analyses, geographic analyses, tests for spatial randomness and cluster analyses, which are common analysis styles in all sciences and can be perfectly put to use in the field we call "digital anthropology".The results show that the Waishengren varied in their practices and their social structure from North to South and through time.
According to our interpretation of the data, Waishengren assimilated and potentially even merged into existing Holo communities in the South, while in the North, Waishengren developed social distinctions through their epigraphical and funeral practices, reflecting power and their status as Waishengren.The gap between funeral practices of Waishengren and Holo has been overcome in the South, where Waishengren assimilated to Holo styles.The gap has been visually overcome in intermediate regions by moving the coffin underground, and it has been openly preserved in the very North of Taiwan.We conclude that if these epigraphical and funeral practices represent identities and self-conceptions of Waishengren, and any research and theory in the field of Waishengren-studies will have to keep track of regional variation at different geographic levels.
The archive itself is under continuous construction.Beyond the completion of annotation and transcription of tombs and tombstones already collected, the sampling of data has to be oriented into two directions: A systematic collection of, first, Waishengren tombs in order to support the reliability of quantitative interpretation of inner-group comparisons, and second, of Chinese tombs in China and other places of the Chinese diaspora, for a further development of the diaspora-paradigm.At last, the ThakBong data might prove useful to contribute as a research paradigm to the general comparison of funeral practices in the context of migration, for example, when Chinese have to redefine their funeral practices in the context of other cultures.This approach might also be applied to other historical contexts, for example, the massive political migration of Pieds Noirs in France after the independence of Algeria in 1962.
Beside the many trends we observe in the development of tombs in Taiwan, and beyond the observed trend in funeral practices towards an integration of the Waishengren community, the last decade has seen a shift caused by the promotion of cremation and the storage of remains in bone-ash towers.At the same time, we observe that public urban policy is reclaiming cemetery land, removing tombs, and turning land into parks.Our documentation and research approach are thus geared towards an "archeology of the future" as our object will soon become an endangered patrimony; we try to document before it joins the remains of the past.This trend leads us to question our disciplinary attachment.Even though our research definitively belongs to the social sciences, and Waishengren-studies have had a major impact on sociology in Taiwan, the documentation of an endangered patrimony, if not already an object for historians, will soon become a topic of archeology.The digital archive, however, is flexible to handle this shift and the research methods, exemplified here under the label of "digital anthropology" will remain to a large extent the same.Digital documentation, digital anthropology and archeology of the future must, from the beginning, be seen together to obtain a long-standing impact on cultural studies in general, and Waishengren-studies in particular.

Figure 1 :
Figure 1: Gender Ratio through time for Taiwan's Han population, as calculated from the ThakBong tombstone data.

Figure 2 :
Figure 2: The places where the estimated 576 Waishengren surnamed Gu3 (古) predominantly settled are randomly scattered over the island.The right graph indicates a statistically significant randomness.

Figure 3 :
Figure 3: The places where the estimated 444 Waishengren surnamed Tu2 (涂) predominantly settled cluster in the North-West of the island.The right graph indicates a statistically significant clustering.

Figure 4 :
Figure 4: Tombs of Waishengren soldiers in a) Jinmenxian, Lieyuxiang, with a rectangular tombstone and red characters, the placename to the right and date to the left b) Jinmenxian, Jinhuzhen, written in Mandarin and English from left to right, characters are only carved and c) Taidongxian, Beinanxiang with a top-rounded tombstone, folkloric decorations, golden characters, a photo and candles, the place written in topic-position and the date to the right.The tombstones in b) replace tombstones that had been written in Chinese, red, top-to-bottom on top-rounded tombstones.Photos by the authors.

Figure 5 :
Figure 5: Tombs of Waishengren between tradition, assimilation and invention: a) in Gaoshu, Pingdung, preserving original practices from Wenzhou, China, b) in Mituo, Kaohsiung, assimilating local Holo tomb styles and c) elitist tombs in Daya, Taizhong, opposing local tomb styles through partially invented forms and inscriptions.Photos by the authors.

Figure 6 :
Figure 6: Tombs of Waishengren of different religious orientations: a) Christian tomb inTaipei, with a traditional tomb form as found in China, but not in Taiwan b) a Muslim tomb in Taipei, which follows the rules according to which Muslims have to be buried in the ground, and a Buddhist tombstone in Yanchao, Kaohsiung, covering probably an urn.Photos by the authors.

Figure 7 :
Figure 7: Rectangular tombstone forms of Waishengren and Holo through time and space: 7a) a time-line analysis, 7b) a histogram where Waishengren and Holo are compared for different administrative units and 7c) a geo-temporal analysis of how the assimilation developed.

Figure 7c )
Figure 7c) finally tries to depict the temporal and spatial pattern of the assimilation of Holo features by Waishengren for each sampled graveyard.Green are Holo and blue Waishengren.The darker colors represent top-rounded tombstones, the light colors represent rectangular tombstones.The dark gray contour shows assimilation of the Holo style before 1980 and the light gray contour shows assimilation of the Holo style after 1980.The assimilation of the top-rounded tombstones was thus completed to the south of the line Taizhong-Xindian-Hualian before 1980.Further north, there is less assimilation by Waishengren and instead Holo assimilated the Waishengren style.

Figure 10 :
Figure 10: Origin of Waishengren according to the ThakBong sample

Figure 11 :
Figure 11: An example of a Waishegren tomb, where the coffin is stored above the ground and the tombstone is rectangular.The tomb is located at the Ningbo Tongxianhui Muyuan in Taipei.Photo by the authors.

Figure 12 :
Figure 12: Clustering 50 randomly chosen Waishengren tombs suggests a classification ofWaishengren tombs in two or three classes.Below the tree-structure, the tomb identification and the date of tomb erection are listed in addition to the variables used for clustering.The labels "covered", "above" and "inside" describe the position of the coffin, "tr" means top-rounded tombstone, "4c" means rectangular tombstone and "loc" means placename in topic position.

Table 1 :
The percentage of Waishengren and their gender ratio in 1957