Sacred Geometries in Contemporary Pilgrimages


Robert H. Stoddard

Department of Geography

University of Nebraska

Lincoln, Nebraska, USA


Pilgrimage flows, generating complex spatial patterns, usually defy uni-dimensional explanations. Nevertheless, observers who attempt to capture the core components of pilgrim behavior may seek understanding primarily through the patterns of sacred geometries. This paper examines the merits of such as they pertain to perceptions of potential pilgrims journeying to a set of sacred sites in the Kathmandu Valley of Nepal.

According to holy texts and religious scholars, the religious geography of the Kathmandu Valley coincides with an enveloping mandala, with temples marking key positions. This geometrical configuration is reinforced annually when pilgrims undertake a forty-kilometer circuit while worshipping sequentially at a set of four Narayan temples within a single day. These locational arrangements and pilgrimages substantiate the contention that cosmological geometries are important.

Complexity in understanding pilgrimage paths is added, however, when the perceptions of potential pilgrims are examined. According to the responses of sampled persons residing in the Kathmandu Valley, very few persons actually envision an encompass ing mandala as the basis for temple locations and, hence, for pilgrimage routes. Although these results can be challenged, they do reveal complexity in the spatial domain of pilgrimages.

- - -

Pilgrimage flows, generating complex spatial patterns, usually defy simple explanations. Nevertheless, for the purpose of formulating generalizations that aid is establishing relationships among phenomena, scholars may focus on only a few factors, which, for the topic at hand, seem to explain the spatial distribution of sacred sites and pilgrimage patterns. Some scholars emphasize explanations based on the distribution of topographic features -- such as mountain tops, the confluences of streams, an d hot springs -- because they are regarded as holy by various religious groups. Other explanations are expressed in terms of historic events that make a site holy, such as the places of birth and martyrdom of religious leaders.

Yet another locational explanation originates with cosmological concepts. That is, holy places are not regarded as primarily the result of unique historical events nor are they evidenced just by distinct earth features. Instead, they occur because the arrangement of sanctified space corresponds to the perceived spatial organization -- or cosmology -- of the universe. It is this latter emphasis that I am discussing here. While recognizing the complexity of pilgrimages, I want to focus especially on the question of whether the pat-terns of sacred geometries can help in understanding pilgrimage behavior.

The Mandala as a Geographic Model for Sacred Places

The importance of the mandala as a geographic model for sacred space is well known (especially to the participants of this conference). For many societies throughout history, the mandala or yantra has been regarded as a graphic representation -- or map -- of the cosmos. This geometric projection of the universe consists of a pattern that surrounds and shapes the lives of inhabitants. By representing characteristics of the universe, the diagram structures religious activities, including movements to specific places.

Certainly in South Asia, Hindu and Buddhist cosmological ideas have influenced the arrangement of the built environment and the associated activities for centuries. Early texts contain detailed rules for designing virtually all objects, including t he form of temples and the lay-out of towns. In Nepal, for example, visual evidence of the mandala as a pervasive model in the design of the built environment is common, especially in Newar settlements. At the scale of an individual house, its siting, ori entation, and relative proportions are based on the same geometric principles as expressed in a mandala.

At a different scale, the arrangement of buildings within a town may also manifest a mandala. The situation has been stated well by Niels Gutschow (1973, p. 43):

One of the models Hinduism employs for the sanctification of space is configurations of shrines or temples dedicated to a particular, systematic group of gods or goddesses. Such shrines cannot be fully understood unless interpreted with reference t o the entire system of which they form a part.

Such systems had to be imposed onto a town that existed already. Hence, the sites of temples or piths are of course conditioned to factors of topography; we do not find the ideal, symmetrical pattern of a well-executed yantra or ma ndala. The variations that actual locations will show do not, however, justify doubt in the principles underlying the spatial arrangement.

At yet another scale, the importance of sacred geometries is clear in Gutschow's discussion about sets of Matrika shrines that encircle inhabitants' sacred space (1977, p. 310):

In the case of Kathmandu, the first of these circles[,] by and large[,] surrounds the city, the second encompasses the valley, and the third at times transcends the boundaries of the valley proper to include places traditionally under the spiritual . . . domination of Newar kings.

Within such a setting, the question arises about the effects of these cosmological concepts on patterns of behavior today. That is, to what degree do beliefs in the sacred geometry of the mandala affect the flows of pilgrims in Nepal today?

Evaluating the Cosmological Model in Nepal

Prior to examining pilgrim patterns, we should first look at the actual configuration of temples and pilgrimage destinations. It is difficult to make an unequivocal evaluation of their arrangements. Most observes agree that the four main stupas of Patan, with one located in each of the four cardinal directions at the edge of the city, do constitute a purposeful pattern modeled after the mandala. The sacred geometry of Bhaktapur, however, is less clear. 0n the one hand, the city -- which is the unio n of two parts, each built around a royal square but at different times -- does not, according to some observers, display the regularity of a mandala. On the other hand, the city does reflect, as stated by Pieper (1973, p. 64), "a superimposed 'iconograph ic' mandala." Certainly the arrangements of various sets of temples and their accompanying activity areas preserve the structure of the sacred model. Furthermore, the circulation of worshippers and religious celebrations reflect the organization of a mand ala because the shrines to the eight Matrikas (or other protecting dikpalas) that dwell around the city are located at the four cardinal and four intermediate directions.

At yet another scale, we can look at the locations of features within the entire Kathmandu Valley. The Valley, which is circular, with a diameter of approximately 25 kilometers, is a fairly discrete region because of the surrounding mountains. The location of several temples within this region may be best understood when they are considered as members of a set, with each position corresponding to a part of a Valley-sized mandala.

Such an arrangement would resemble the sacred geography of certain mountainous regions, as explained by Grapard (1982). It was in this context that I attempted to discovered the degree to which residents of the Valley actually perceived an enveloping m andala -- one that determined the location of specific temples, and hence the destination of contemporary pilgrimages.

I commenced my investigation with three questions:

(1) Are there sets of sacred sites that apply to the Valley; that is, are there several places that constitute parts of an over-all unit of sacred space? If "yes," then:

(2) Are the sacred sites located in a geometric pattern; that is, does each place in a set have spatial significance in terms of an over-all religious design? If "yes,' then:

(3) Do most persons living in the Valley perceive specific sacred places as parts of a set which has meaning for the Valley as a whole?

To answer the first two questions, I consulted several experts and relevant publications. The experts were local persons who have studied the ancient religious texts and have specialized in understanding Hindu-Buddhist Tantricism. The publications I searched for answers were ones readily available in Kathmandu and written in English.

According to these sources, the answer to the first question seems clearly to be that, indeed, several religious sites do exist as sets. The experts and various publications generally agreed with what has been summarized in the monumental inventory of cultural features in the Valley, namely, the Protective Inventory (Prusha, 1975). The Inventory lists several religious sets, but I am concerned with only three sets, specifically, the Char Ganesh, the Char Narayan, an the Ashta Matrikas .

The answer to the second question about whether the sets of sites form a meaningful pattern is not a clear-cut "yes". The lack of an unequivocal affirmation is partly because there is not complete agreement about which Ganesh and Narayan shrines ar e members of the two sets of four. The Ganesh shrines and their locations according to one page in the Protective Inventory are Surya Vinayak (south of Bhaktapur), Jal Vinayak (at the Chobar gorge), Rakta or Chandra Vinayak, and Karya Vinayak (Figu re 1). However, elsewhere in the same publication, Ashok Vinayak in Kathmandu replaces Karya Vinayak. And, in one of Gutschow's publications, he lists a Ganesh shrine in Sankhu instead of Rakta Vinayak.

Likewise, the four Vishnu shrines are usually, but not always, given as Changu Narayan (southwest of Sankhu), Ichangu Narayan (near Holchok), Sekh Narayan (near Pharping), and Bisankhu Narayan (midway between Lubhu and Godavari). As exceptions, one publication states that Buddha Nilkantha is even more important than these other four Vishnu temples; and Gutschow and Auer include a sixth one at Macchegaon.

With this diversity in enumerating which places actually constitute the set, it becomes a little difficult to discern whether either set does form a regular pattern that conforms to a Valley-size mandala. Generally the sources that commented on the geography of religious sites did agree with the Gutschow thesis: that sets of sites are theoretically positioned in the Valley in the cardinal directions. Nevertheless, they recognize that topographic features and ancient religious structures have modifi ed a perfect north-south and east-west orientation, and these locational deviations from the ideal make it difficult to determine visually which of the conflicting sites are most likely members of the set. This is especially true in the case of the Ganesh shrines. However, there was greater unanimity about the four Narayan temples; and when Buddha Nilkantha and Macchegaon are omitted, the remaining four do display spatial regularity within the Valley.

When the question about whether a set of eight sites of the Ashta Matrikas forms a meaningful pattern, the answer seems to be mostly negative. This conclusion results primarily because of very limited knowledge about any set that pertains to the Va lley as a whole. Only one of the published references referred to such for the Valley as a whole, and several of the interviewed experts stated that no set of sacred sites exists for the entire Valley. Thus, even though it is well known that Ashta Matrika s surround each of the three major cities, knowledge about a set of dikpalas (or chetrapalas) that protects the entire Kathmandu Valley is very rare.

The answer to the second question about the spatial regularity of sacred sites in the Valley, therefore, seems to be a conditional "yes". There is general agreement about the specific members of the Narayan set, and their locations within the Valle y do, indeed, appear to form a regular pattern. There is less agreement about which four of five major Ganesh shrines constitute the true set, and none of the potential sets forms a pattern that appears obviously regular to an outsider. Thirdly, although the eight Matrika sites may form a well-oriented pattern, as shown by the small map published by Gutschow and Bajracharya (1977, p.4), this information is apparently very esoteric.

Perceptions of Sacred Sites in the Kathmandu Valley

The third research question dealt with how inhabitants perceive the geography of sacred sites in the Valley. Information about perceptions was acquired by asking selected villagers and townspeople about the major Ganesh and Narayan shrines and abou t sites of the Ashta Matrikas. My assistant* and I went to 25 sites scattered throughout the Valley during a month's period to interview 114 lay persons. A heterogeneous sample of persons was contacted, with the respondents representing a variety of ages (Table 1), schooling levels (Table 2), and occupations (Table 3). Respondents were asked to name the four main Ganesh shrines and then the four Narayan sites, to describe the locations of all those sites, and to show their locations on a large-scale map. Then they were asked why the sites are located as they are. Also, persons were asked about the existence of Matrikas for the Valley and for their own localities.

The answer to the question about the perceived sacred geography of the Kathmandu Valley is based on their responses.. First, let's note the extent to which respondents identified sets of Ganesh shrines (Table 4). From Column "a" of Table 4, it appe ars that the interviewed laypersons -- like the experts -- were uncertain about which of five Ganesh shrines constitute the set of four that belong together. This uncertainty is evidenced by the gradual decrease from 92 to 50 in the frequency the top five places were named -- rather than a sharp break after the mentioned top four.

In contrast, respondents were more certain which four Narayan sites form a distinct set, as revealed by the fairly obvious break in frequencies between the fourth and fifth listed shrines (Table 5a).

It occurred to me that persons might perceive a spatial regularity in sacred places even though they might not give such an explanation verbally. In an attempt to detect whether respondents might visualize a mandala-like pattern in the positions of Ganesh and Narayan shrines, we asked the interviewed persons to show the locations of these shrines by placing stones on a large cloth map we placed on the ground. This base map included the Valley rim, along with a few of its demarcating peaks, all the main streams, the major surfaced roads, and a few landmarks -- but, of course, none of the shrines that were the focus of our inquiry. The responses are summarized here by both a frequency table and a map. Column "b" of Table 4 shows the number of respond ents who mapped, as well as named, each of the five main Ganesh sites. It is evident that a strong majority of those who identified specific shrines were willing to indicate where they thought the Ganesh sites are located.

All of the locations marked on our map of the Valley by respondents have been transferred to the map shown here (Figure 2). This technique does not display the locational perception of each individual, but it provides a clue to collective concepts. The general impression conveyed by this map is, from my judgment, not one of greater regularity than displayed by the actual sites. That is, if respondents truly envisioned the temples situated according to the regularities of the mandala, then we might expect them to place stones on the map in a more regular arrangement than the true geographic positions. Instead, the dots seem to cluster around the actual sites, especially the best known ones.

The same procedures was followed for the Narayan shrines. Column "b" of Table 5 again signifies that a large majority of those who named a particular Narayan shrine also indicated on our map their answers about its position in the Valley. Responden ts' perceptions about the locations of Narayan shrines are shown in Figure 3. The collective perceptions indicate some regularity, but this could well be because the actual sites are, in fact, situated in a fairly regular pattern. Consequently, the result s may reflect good mapping skills rather than a tendency to place the stones according to a perceived mandala pattern.

Since even the experts are uncertain about the existence of Ashta Matrikas for the Valley, it comes as no surprise that respondents also said they did not know about such. In a few villages, persons identified a particular place which did, in fact, correspond with one of the Valley Matrikas listed by Gutschow and Bajracharya (1977); but they invariably considered it as a place of importance only for the immediate locality and not as a member of a set for the entire Valley. On the basis of the answe rs provided, one must conclude that there is no common perception among the residents of the Kathmandu Valley about the existence of a set of Matrikas protecting the entire Valley.

In reply to our question about why Ganesh and Narayan shrines are located as they are, only two persons pointed out that the sites are located in the four directions; two others regarded the shrines as protectors for the Valley; and three interview ees specified that the four shrines were situated to serve the entire Valley. Four persons declared that the position of Surya Vinayak in the east is significant, but they did not mention the other three as having similar locational importance. Another th irteen told a specific story as the reason for a particular location being sacred. Approximately 80 per cent of those who answered this question, though, did not provide any reason for the spatial distribution of sacred sites.

Before forming any conclusions, I should mention some potential problems with these data. For one, respondents were not selected by a statistically random procedure. That is, no prescribed method for using a sampling frame was employed, and thus th e respondents cannot be regarded necessarily as an unbiased subset from the general Valley population. However, if there is any bias, I suspect it is that persons having more information about places other than their immediate vicinity were over-represent ed. For example, a higher percentage of school teachers were interviewed than their proportion of the total population (see Table 3), thus producing a sample of persons who undoubtedly leave their home village more frequently than do farmers. It is probab ly safe to assume that persons not chosen for interviewing were even less aware of religious places outside their immediate locality. Therefore, this possible source of error would not jeopardize a negative conclusion about perceptions of patterns in the locations of' Valley sites.

Secondly, respondents may not have given correct information. It is recognized universally that, in any technique that depends on the cooperation of others, people may deliberately chose to misled the interviewer. However, the percentage of persons in the Kathmandu Valley who were willing to answer questions compares very favorably with other parts of the world and indicates a positive attitude about giving valid answers.

Respondents may have given incorrect information unintentionally. This problem could arise from faulty communication. Under the best of conditions it is possible that an interviewer and interviewee misunderstand what the other person is trying to s ay. The possibility of such misunderstanding is increased if the information must be communicated through a third person, especially when the conversation must be translated into another language. Fortunately many of the questions and answers were fairly factual, so these data were probably communicated satisfactorily; but the attempt to learn about respondents' beliefs about regularity in religious sites undoubtedly was less successful.

Incorrect information could occur from memory lapses or similar mistakes. This situation is illustrated by those who were not able to remember names or places or to use the map. Some persons knew where a Ganesh or Narayan temple was located but cou ld not remember its name. Others knew a name but forgot the location. And, of course, people vary in their ability to understand and communicate an isomorphic relationship between earth locations and map positions. Analysis of the data, however, does not suggest any major sources of limited knowledge. Columns "c" and "d" of Table 4 provide information about the level of mapped replies by those who visited each Ganesh shrine within the previous 12 months. The percentage of those who had visited the sites a nd also mapped them was high -- not only for the frequently visited Surya Vinayak and Ashok Vinayak shrines but for the other three also. Although the number of respondents who visited the Narayan shrines was generally lower than the Ganesh shrines (Table 5a and 5b), the percentage of visitors who also mapped the shrine locations remained high .

My attempt to discover whether physical proximity to shrines made a significant difference in their being named and mapped is summarized in Table 6 and Figure 4. The latter shows the location of places where clusters of interviews were conducted. A s I mentioned above, these were distributed throughout the Valley, so in the analytical stage of my work I roughly grouped five interview places near Changu Narayan, five near Surya Vinayak, and five in each of the southwest, northwest, and in the central -south areas. According to these groupings in Table 6, there is a decline in familiarity with Changu Narayan and Surya Vinayak with increased intervening distance. For example, 79 percent of those residing near Changu Narayan named and mapped this shrine whereas only 45 percent of those in the northwest did so. Likewise, 94 percent of respondents living near Surya Vinayak named it along with Changu Narayan, but only 55 percent of those in the northwest achieved this level of identification. Nevertheless, in spite of these locational differences, the overall level of performance was high for these selected tasks.

Other cross-tabulations did not expose any significant differences in success rates. It is true that females did not name and map selected sites as well as males (Table 7), but this is not a surprising outcome, given the greater mobility of males.< /P>

Any one, or a combination, of these conditions may have caused errors in the data. However, if the assumption is made that whatever errors did occur are not systematically large and of a single type, then results can be viewed as a satisfactory ans wer to the question about inhabitants' perceptions of' sacred places.

Assuming that the data are reasonably accurate, we return to the third question: Do most persons living in the Valley perceive specific sacred places as parts of a set which has meaning for the Valley as a whole? The answer seems to be mostly negat ive.

In the case of the Matrikas, this conclusion may result from at least two character-istics of Tantricism. One characteristic is its secretive nature, which deliberately limits a wide diffusion of knowledge about beliefs and practices. Another chara cteristic is the manner in which knowledge is passed from guru to disciple rather than through well-publicized doctrines, which means that certain sites may have meaning for one group of believers but not the same significance for followers of another tea cher. Therefore, it is safe to conclude that the arrangement of any religious sites which may be associated with the eight Matrikas is not commonly perceived as representing a Valley-wide yantra or similar representation of a cosmic design.

For the Ganesh and Narayan sites, the conclusion is partly "yes" but mostly "no". On the positive side, most respondents did consider four specific shrines as belonging to a set (i.e., one set for Ganesh and one for Narayan), and each set was ident ified as distinct. Further evidence that the four shrines are regarded as members of a unique set was acquired when a couple dozen respondents reported that, during the last ten years, they had made the special circular pilgrimage to all four Narayan shri nes.

On the negative side is the fact that, although residents of the Valley may view the various shrines as belonging together in one of the two sets, very few respondents gave any reason for the locations of the shrines. In contrast to the experts, wh o usually related the positions of the shrines to the cardinal directions in the Valley, the lay persons seldom mentioned any over-all pattern. Many respondents did answer queries about local shrines around their own village or town in terms of' protectiv e positions, but they rarely applied this same kind of locational model to the sets of shrines in the entire Kathmandu Valley. Consequently, it seems safe to conclude that, although inhabitants regard spatial regularity as very important in the sacred spa ce of their local vicinity, they do not necessarily perceive a similar geography of sacred sites at the scale of the Valley.


This paper focuses on whether concepts of cosmic geometries can help us understand the complexity in pilgrimages. In some cases, cosmological concepts may be very helpful. For example, each year at the time of Haribodhini Ekadasi, pilgrims make a o ne-day 40-kilometer journey to a set of four Narayan shrines in the Kathmandu Valley. By their movement, pilgrims delineate a mandala-like pattern because the shrines are arranged according to a cosmic design.

However, this case cannot necessarily be accepted as a universal model for explaining pilgrimage patterns. There are numerous pilgrimages where the routes and destinations have not been expressed in terms of cosmic geometries. Furthermore, even in this situation, there is evidence -- as I have shown -- that many pilgrims are not conscious-ly aware of an encompassing cosmic geometry governing shrine locations. This does not deny that early traditions may have developed from cosmological concepts, an d that today, such origins are not known by all laypersons. If such is the case, then maybe pilgrimages routes can be associated with cosmic designs, even though pilgrims may not be aware of such origins for the paths they follow.

Nevertheless, this study indicates that knowing about the cosmological concepts of a particular religious group does not necessarily simplify our understanding of pilgrimage patterns. Indeed, the complexity of pilgrimage remains as a major challeng e to those who seek to understand this fascinating phenomenon.

Return to Submitted Abstracts List Page

Return to Pilgrimage Workshop Home Page