Introduction

This paper was presented at the workshop on ‘War and Wartime in Visual Representations’ held at the University of California, Berkeley on August 23-25, 2006. It is in draft form, but I am willing to make it available for discussion among the scholarly community. It is part of the attempt made by the participants to the “Virtual Shanghai” project to explore new ways of writing historical narratives. The present paper includes images that pertain to its topic, but it is also related to parallel visual narratives, one based on photographs, the other based on maps. These two visual narratives are currently still on the former "old" Virtual Shanghai platform, pending further programming.

Photography figures prominently among the visual sources that have enriched the palette of materials used by historians over the last decade. Born with the industrial age, photography rapidly evolved from a practice geared toward individual portraiture or landscape by photo studios or skilled amateurs to a major mode of communication in its own right. The role of photography was magnified by its early association with the press. By the turn of the twentieth century, photography was rapidly displacing the previous modes of illustration in periodicals. In 1900, the journal L’Illustration already drew 41% of its images from photographs. By 1914, their share had increased to 78%.¹ The growing impact of photography was also related to its use on the battlefield. War, and especially W.W.I., gave a major impetus to the rise of photography as the primary medium to report “on reality” from the front lines back home. Armies were not slow to set up photographic services to cover on-going conflicts. Individual periodicals would also send reporters-photographers to be joined soon by press agencies specializing in providing photographs of major events all over the planet. Used for propaganda to boost morale as much as for information, photographs thus became a central element of the emerging mass media.²

War photography definitely ranks as the genre par excellence that attempts to document in visual form for the readership far away from the frontlines of the battlefield – at least until W.W.II. – one of the most brutal manifestations of man. As photography became more popular, along with the emergence of professional press agencies, it recorded in increasing number the ravages of modern warfare on both soldiers and civilians. War photography has become a “genre” in its own right, producing emblematic images of modern conflicts.³ Altogether, however, this apparent abundance must be balanced with various factors such as a focus on certain aspects of war and neglect of others, frequent absence of proper labeling, and of course important losses in the course of time. Nevertheless, photography offers a wide field to be explored, even if it raises particular challenges for historical research.

This paper aims to explore how historians can use the photographs taken during a war – the Sino-Japanese conflict in Shanghai in late 1937 – as a source to understand certain processes, especially the impact of war on ordinary people, and as a way to both recount the events themselves and to contribute to a reasoned memory of these events. The first part of the paper deals with the general issue of using photographs as a historical source for the modern historian. I argue that photograph offer a privileged medium that seemingly allows the historian to reach out toward its subject matter while it also takes him/her onto a largely undetermined field, both on account of “knowing” the past and of “telling” history. The second part of the paper presents a narrative of the experience of refugees in Shanghai based on the use of photographs. It claims that photographs open windows on aspects that were left off the record or allow a better understanding of the tragic fate of the refugee population. While images are presented within the text to visualize this perception of the past, their illustrative function can hardly be dismissed. Yet I also propose parallel narratives in the form of visual and map narratives not to supplement the textual narrative, but as actual alternative readings of the refugee experience.

Notes about the use of photography in history

“[The] meanings of photographs emerge from cultural experiences; they are ideological reproductions of reality and viewed as realizations of an objective truth. They are also a source of multiple collective memories and therefore components of history as a creative process.”⁴ This quote somehow sums up the dilemma of historians when dealing with non textual records. Images, visual materials, etc. are not, of course, something new on the academic planet. Even among historians, images have provided a rich staple to feed historical inquiry. For the most part, however, this has been the realm of either historians of pre-contemporary periods (from Ancient to Modern history) or, quite naturally, of art historians. Beyond this specific realm that already offers a lot of insight, methods, and practice in the use of visual materials, the “visual” has taken up a new dimension with the development of “visual studies” built around the concept of “visual culture”. Visual studies actually draw on a wide array of disciplines and offer various ways into the study of visual sources.⁵

I shall not discuss this field here. First, because this is beyond my competence. Second, because a large part of the theories, concepts and even methods are not necessarily relevant for the purpose of a social historian. This is not meant to dismiss visual studies. I have found very illuminating to peruse many of the works produced on visual culture, especially given my own interest in using the two “faces of the coin” any visual source offers. My analysis here will be limited to one kind of visual material, photography. Photographs are no stranger to historians. They have been used quite often in books, sometimes to identify – visually – a particular figure, or a particular place. More often than not, photographs appear as an illustrative companion to text, never or rarely disjointed from the broad narrative presented, but yet often not very close to an actual reference in the text and, above all, seldom used as a source of information feeding in the narrative itself.⁶ What we are left with, in most cases, is the use of photographs as illustrations, an illustration meant to convey a “sense of things past” by providing a tiny visual window into that past.

This is much below what photographs have to offer. One can definitely go beyond this level and take photographs seriously as a source in itself. A type of source, however, that raises new challenges. When using photographs, historians are caught in a tension – the source/narrative – tension. Images can catch the eye of an audience and create a sense of proximity that is engaging and therefore effective. “In that sense, photographs also serve the imagination of historians and feed their creative instincts.”⁷ They also confront them with temptations about their various possible uses:

• to see the past → as a source, as if photography could relate to us a

concrete view of ‘how things were’

• to illustrate the past → as a mere adjunct to a textual narrative, though

in most cases the narrative could do without the images (e.g. portrait)

• to visualize the past → to reconstruct a visual perception of the past and

to show it to the reader (from “systematic sample” to “visual narrative”)

This is, of course, a crude and probably incomplete view of how historians can use photographs in their work, but this is meant to place a certain emphasis on the major ways in which, I believe, we can use photographs as historians. Probably, for some time to come, the use of photographs as illustrations will remain the dominant use. This is not a bad thing. These images do not distort the historical narrative and they enliven the long text through which most of us entertain our readers and train our students. One should also note here that photographs have been used as the core material of a “history of”. Yet it is usually in the form of albums where conversely text is reduced to a minimum or altogether absent. There is rarely any attempt to construct a signifying narrative through the selected photographs beyond a chronological or topical line.⁸

To come back to the possible uses of photographs, my interest lies primarily in the two “non illustrative” functions, namely “to see the past” and “to visualize the past”. When a photograph comes into the hands of a serious historian – serious enough to give it some credit as a legitimate source – it entails an immediate quest for the form of source criticism historians apply to textual sources. In many cases, there will be only partial answers to the most basic questions:

• Where?

• When?

• By whom?

• Why?

• How?

• What for?

Before even looking into the content of the photograph itself, this is what a historian should know about a photograph. This is after all the same kind of questions we raise about written sources, except that both for historical reasons (photographs have not been collected and processed in the same systematic way as textual records by libraries or archives) and reasons related to the nature of photography (it comes as a fixed and unique image), historians cannot rely on the same set of methods they were trained into.

Photography comes in as a bunch of built-in contradictions. It appears to offer a slice of “true reality”, of a moment that did exist – the camera recorded it – and yet what we see depends on the conditions under which the photograph was taken. Moreover, a photograph does not come as a set of items one can examine and deconstruct in the same way historians examine a text (from its source, its language, its support, etc.). A photograph is one and a whole. This is not to say there are no specific tools to “deconstruct” a photograph (art historians have a complete tool box), but the purpose a historian has in mind only partly overlaps with the approach of a photograph as a object of art.

A frequent difficulty in using photographs lies in their lack of proper identification. This is due to the fact that photographs were not considered as a significant source in itself (rather a record of the past that may be used for exhibitions). In many cases, photographs came in from various sources, were put into boxes in bulk – sometimes for decades – and sorted out at a time when there was nobody left that could provide information data. Photographs were also taken as an adjunct to reportages: “photographs were used to position targets of depiction within a larger story of Nazi atrocity”.⁹ This holds true for other “stories”. This difficulty is compounded by the fact that many pictures, especially after the popularization of the camera in the early twentieth century, were produced by all kinds of individuals, not just professionals. In my own experience of Shanghai historical photographs, most were taken by “Mr. Anonymous”. We are left, therefore, with a very rich visual material, but very few indications about its nature, source, conditions of production, purpose, etc.

One way to get around this problem is to work on a specific collection of photographs, produced either by an individual photographer or by an institution (commissioning professional photographers) or press agency/newspaper. Depending on the degree of rigor of the photographer, the historian will avail himself of a better knowledge of the circumstances under which the photographs were produced. Yet, a whole array of difficulties remains. First, the photographer may not have left a proper or detailed record of what, when and where the pictures were taken. One can be in Peking (but anywhere in Peking) at any time between 1933 and 1946 and be shown unidentified places, artifacts, people, etc.¹⁰ The other issue with identified individual photographers is that most of them wiere professional photographers. There are of course numerous examples of quasi professional “amateur” photographing, especially in the nineteenth century (e.g. diplomats, missionaries, etc.). It probably creates an intermediate category between sheer amateurs and professionals (e.g. such examples as Sidney D. Gamble or Father Joseph de Revières, see infra). On the whole, however, the development of amateur photographing in the twentieth century entailed the production of much less focused and “coherent” works. It also means a large constellation of disjointed images.

Professional photography, therefore, may appear as a more secure ground to work on for the historian, and yet this is an illusion because it implies that picture-taking will be highly conditioned by the purpose photographers have, a major one being to make a living, either by selling their pictures to a certain audience or to magazines (or by the periodical’s editorial line in the case of employed photographers). Professional picture-taking results from an equation that integrate personal inclination (and subjectivity), purpose (sale, magazine), intended audience, and external conditions (social context, technological constraints). Skilled amateurs may have a greater leeway as immediate purpose and targeted audience may be quite vague, but the subjectivity will be there too. In other words, what one gains in better background knowledge, one loses on the side of “spontaneity”.

Before moving toward the discussion of the photograph itself, another point deserves our attention. Photographs were produced in increasing numbers between the moment of their invention and the mid-20^th century. In fact, when the camera started to become a consumer good after W.W. I, millions of photographs were taken by individuals, studios, press agencies and newspapers, institutions, etc. And yet most of these pictures were lost, as happens too with all our other sources. Probably more in the case of photographs, leaving an advantage to those produced by institutions and press-related organs. Our “vision of the past” will require correcting lenses to make up for this imbalance between “public” (including here newspapers) and “private”. In other words, if photographs are thin slices of the past, the stack of slices we have has been reduced to a trickle from the original flow of images. In the case of China, this issue is compounded by additional obstacles related to the historical circumstances the country went through (war), institutional instability, lack of preservation and, as far as the PRC is concerned, utmost difficulties in accessing original collections.

What this comes to, eventually, is the realization that photographs will never match the range and depth of issues that textual records, especially archives, are able to cover. Despite the variety of circumstances under which photographs were produced, at least until W.W. II, photographs were never produced in the same way as written documents were by government agencies – say by the Bureau of Social Affairs or the Ministry of Interior (though Homeland Security has definitely improved the records, especially for portraits of individuals). Photography was a mere accessory to administration, business management, etc. There was never anything close to the systematic production of written documents (or rarely so: French photographic coverage of historical monuments in the late 19^th century).¹¹ A major consequence is the loss of the kind of traceability a historian is able to recover through the textual records.

Photographs will for ever be just that: shots into the past. Individual shots, shots in series, shots in sequence, shots by one single photographer, shots that offer more or less density and granularity on a place, an event, or on people. But shots that bring to the surface a closer sense of experience (what it was like), shots that bring to light aspects of the past that the textual documents failed to record or pushed back into the shades. A unique quality of photography is its immediacy. Its conditions of production differ quite radically from that of textual records. On the one hand, any text is always the result of a construction made by its author, be it a technical report on road repair, a survey of the use of drugs or a police account of a crime, etc. Such elaborations may go through various preliminary stages, use different techniques (data collection, interviews, etc.) and involve different “hands” before a text is written. There is much less distance – both temporal and spatial – between a photographer and what he sees than between an author and what he narrates.

On the other hand, without underestimating the cognitive dimension of picture-taking – I am aware, as noted previously, that the photographer does make choices, that his/her images are not a guarantee of more “objectivity” – yet the time for elaboration is far more limited, especially in stressful situations like war photography. Moreover, the photographer can hardly change what is in front of the camera’s lenses. What we gain in more immediacy and more direct “contact” with the past, however, we also lose in terms of depth. The historian gets a framed still image of a split-second moment of the past at a given location. And digging into the surface of a single photograph definitely raises a new challenge as more often than not we shall end up on the other blank side of the photograph. Once in a while, some tiny help will come in the form of scribbled words on the back, but even this may be misleading, even with professional photographers.¹² In fact, as I shall argue below, the blank side is entirely up to the historian. This is his/her own work ground.

A photograph is like a coin: it has two faces. On one side is what I see, the image that offers itself to view and interpretation. The other face is more metaphorical. I do not mean here any annotation or whatever may appear on the verso of the photographs (though this should not be discounted). The other side is what I know about the photograph (what, when, where, by whom, etc.). Having all the relevant data is a considerable advantage to interpret what the photograph shows, what the photograph tells us. Yet it is not the whole story, as I try to explain below.¹³

For the historian, to read a photograph is a challenge to his/her habits. At a preliminary level, a photograph will be read as a whole. It will convey an immediate message or meaning. It is not something that comes through the reading of words that progressively build into a sentence and eventually produce a meaning. A photograph comes as a “bundle” of meaning(s). One sees a photograph before reading it. I shall not deny here that our eyes are not mere cameras recording mechanically what comes within their range of vision. They are connected to a mind that will add “reading” to “viewing”, including injecting meaning onto the image, projecting questions about it, and moving toward more systematic deciphering. The range is wide, from a simple view like a portrait (nothing more than a face, a hair/beard style, a collar, a hat and the “identified-unidentified” dualism) to complex scenes (groups, streets, landscape, etc.). There are many ways to “deconstruct” a photograph depending on one’s perspective and focus (the image [as object, as event, as representation, etc.]? the author?). One can delve on various disciplines as “visual studies” tell us. There are tools in there the historian can draw on to guide his/her steps at this level.

The other face of the coin is as important as its “positive” side, especially when one deals with photographs that come in with little or no information. In fact, any reading of a photograph is made at the point of intersection between “what I see” and “what I know”.¹⁴ This holds true of other sources, but with images, especially when they come with little background data, one can only rely on pre-existing “knowledge” to read a photograph and give it proper meaning. “What I know” includes both knowledge on the photograph itself and knowledge outside of the photograph. Knowledge on the photograph refers to all information data covered by the usual set of questions spelled out above (what, when, where, etc.), while knowledge outside the photograph means everything one knows about the broader context of the photograph (which may be as wide as “this is in China”, no place, no date). It includes any knowledge that will condition the reading of a photograph. This is what I call metaphorically the other face of the coin, the “blank” side of the photograph, the work ground of the historian.

My own approach as a historian is based on a very simple method. When reading a photograph I start with a systematic itemization of what it shows (and what it does not show). This is meant to “objectify” as much as possible what I see and keep a distance with any information or knowledge I may have about the image. The second consideration is to attempt to pick on any detail that would contribute to a proper identification and interpretation of the scene under scrutiny. A photograph should never be taken lightly. Details are what may help eliminate options and therefore narrow down the topic, time frame, location, etc. shown on the photograph. Even when the view is obvious – e.g. a picture of the Bund – one may be able to date a picture from the buildings shown and not shown (they changed over time at different dates) or the size of the trees (provided one knows what kind of tree and when they were planted). In other words, a rigorous ethnographic or/and archaeological reading is a preliminary and most necessary step for the interpretation of a photograph. This will produce a complete record of identified/unidentified items and layers displayed on a photograph.

It is on the basis of this “ethnographic report” that the historian can move to his/her own work ground because the task at hand will work in two directions. The elements of information available on the photograph – following a thorough verification – as well as the prior knowledge of the historian will change the contrast, sharpen the edge, and highlight parts left in the dark or shadows. Conversely, the “ethnographic report” will serve to challenge or test prior knowledge as well as indicate directions for further research on visible, but unascertained or unidentified aspects/items displayed on the photograph (type of tree, date of plantation, date of building, type of car, etc.). In many cases, this attempt may prove fruitless and hit a wall. Yet for any picture to be used as a source, these are the steps to be taken if one does not want to be caught using a “wrong” image. When using pictures on a large scale, for instance in establishing a database, there is no need to go for such a systematic examination. This would be an endless pursuit. Systematic and close scrutiny only makes sense when a picture is solicited as a source in a visual or textual narrative. Of course, when an image database exists, this work is made easier thanks to the possibilities of comparison between images that deal with the same topics.

From the perspective of using photographs as a source, the historian will meet with basically three modes of picture-taking:

• Posed/Staged photographs: Despite the similarity, the two modes are not quite the same. Posed photographs were mandatory in the early stages of photography when the time of exposition was counted in periods of up to 15 minutes. A photographer could take a picture of a “real life” situation, but that “real life” situation had to be created on purpose with all the participants frozen in immobility for the sake of picture-taking. Another kind of photograph found in nineteenth-century China was staged photography. This is a subtle shift from posed photography, but it entails a fundamental change of perspective. In a staged photograph, the photographer “creates” himself a situation – e.g. pictures of prisoners in cangue, picture of a capital execution – that is not real. It is reinvented, re-enacted for the purpose of picture taking. The issue with such photographs is that they may well faithfully tell a “true story” (provided that all relevant elements are in there) that was impossible to document in a real life situation. The status of such photographs is quite ambiguous for the historian due to their mixing of reality and fiction. To some extent, though, this may be compared to accounts of such events in novels (true description, but invented situation). Generally, given the cost, weight, and skills required, photographs before 1900 were made by a small number of individuals – with little distinction between amateurs and professionals – necessarily carefully choosing their targets – in other words, with a purpose (audience, field of interest) – and excluding situations outside the bounds of available technology.

• Targeted photographs: the second major category of photographs is that produced by professional photographers after the transition to cameras allowing instant snapshots. It is a major category because it can be assumed to have better survived the challenge of time and to provide a wealth of visual documentation to the historian. It is also significant for its quality – professional picture-taking – and its relevance: professional photographers, especially those working for press agencies or newspapers, were expected to cover “events” (strike, fire, accident, etc.). Often too, photographers had their own field of interest and covered a variety of subjects outside their strict professional commitments. In using such photographs, the historian is confronted to two major issues: the first one is that of identification, the second one is the nature of professional photography. The first issue is not easy to solve. While famous photographers and their production are easy to trace (Henri Cartier-Bresson, Robert Capa, etc.) with complete bibliographies, the lesser known ones – most photographers employed in the press – are often impossible to identify and to relate to specific pictures. Even someone like Hedda Morrison, who belatedly came to notoriety, hardly left anything about herself. Yet all photographers share the same approach: they are very selective in their practice of photography. They take what sells, or potentially so, or what their employer expects to sell. This is a target-driven practice of photography that carves out thin slices of time meant to represent far more than what they actually show. Looking back at them after years or decades, the historian is caught in a genuine “reality/representation” dilemma.

• Random photographs: I include under this category all the pictures taken by non-professionals, most of them for a personal purpose (travelers, sojourners, local residents, etc.) rather than for a broader audience. The range also covers “enlightened amateurs” like Sydney D. Gamble – who had a long practice of photography which he used as part of his missionary activities in China, but before all as a personal hobby – or missionary-photographers like Leone Nani, the Jesuit father Joseph de Reviers, or diplomat-photographers like Auguste François.¹⁵ A large part of the stock, however, comes from people who took pictures “on the fly” rather than with a specific frame of mind, ordinary people – especially foreigners – whose eyes were caught by an aspect of life in China (e.g. Karl Kelgenbacher's pictures of wartime Shanghai on Virtual Shanghai). A recurrent issue with such photographs is often the lack of proper data, in particular when family albums end up in the hands of the third or fourth generation. While the bulk of such pictures fail to present much historical interest or are of poor quality, many also provide a “fresh” and less mediated look at what was taken by the camera.

For all three categories, a point needs to be made about the Chinese/foreigners distinction. Obviously, a large part of our stock of images, especially for earlier periods, is made up of pictures taken by foreign visitors or residents, some of them first-timers in China. This process was renewed throughout the 19^th and 20^th centuries, even if the circulation of images brought home various layers of prior “knowledge” to people bound to travel to China. The Chinese/foreigners distinction is also important to keep in mind when thinking about what is shown of China (apart from news photography where nationality loses its relevance). It can be expected that foreigners looked differently at Chinese society, they reacted to things they felt surprising, alien, unusual, etc. What we may be lacking here are collections of photographs by individual and identified Chinese photographers to tell us how they “saw” their own society.¹⁶

What this comes down to is the extreme variety of visual documents photography has left us. Apart from issues of quality, they all look alike – black & white shots of reality – but they were produced by very different types of “picture-takers” with different perspectives. These images offer a very tempting material that seems to materialize before our eyes the ever illusive past the historian is so desperate to grasp in all its dimensions. This material, however, can also prove very illusive and must never be taken as an image of the past. It is to be considered as an image from the past, a document that requires a difficult exercise in reading through the two sides of the image – what it shows and why/how it is showing this. A single image can take one onto several directions.

A visual narrative of the fate of refugees in Shanghai (1937-1938)

In the preceding section, I have been discussing the upstream side of using photographs for the historian. As a source, the historian will build his approach on a triptych – see/experience/know, both for himself (what I see, what I experience, what I know from the photograph) and about the object of the photograph (the object I see, the experience I observe, the knowledge I gain about the object). On that basis, he may also want to share his view/experience/knowledge with his reader(s) not just by writing about or from the photograph, but by showing it to the reader. This entails a new challenge as the use of photograph downstream will be based on a confusion (or concatenation) of usually separate operations in the historical narrative: the written narrative on the one hand, and the original documents (most often as appendices) on the other hand.

A photograph will by its very nature encapsulate both aspects: it seems to provide a guarantee of authenticity (original document) while it is also meant to convey a message with the general historical narrative. In such a use, the historian is confronted to the set of issues scholars in visual studies have pointed out, especially the fact that the historian has no control on how the images he has selected will be read. In fact, he has a certain degree of control through the use of captions (and text if the photograph is discussed in the text). This may even be pushed toward manipulation. The issue is made even more complicated if one uses images that have become “standards” or even “icons” as it will be difficult to detach them from the set of pre-defined meanings they came with.

In its illustrative function, an image is often used to let the reader “see” an object of history. The “know/experience” functions are less prominent.¹⁷ On the other hand, the choice of the picture will definitely increase the arbitrary and subjective dimension of how the past is shown. It is an end-of-the-line selection made out of what history has left of what the original photographer had decided to take. In other words, by doing this we increase by a very large factor what the picture is showing to take it as the (true) representation of a much larger historical moment. This is not an issue since the image is here only as a supplement to the text and since the image makes a limited intervention into the narrative. It does not diminish the need for a careful selection based on a thorough critical review of the original image. Many of us have worked along these lines with photographs, usually with a limited number due to publishing costs, but more extensively in conference presentations.

Some authors have attempted to limit the use of words when using pictures, though these cases are rare.¹⁸ More often than not, pictures come with a lot of words. In working on the history of Shanghai, and more specifically on “war refugees”, I have been struggling with these various issues of which I have tried to present a brief sketch in the preceding section. In the second part of this paper I shall take a concrete case to show how one can operate in using visual documents. Such an experiment is made possible thanks to Internet technologies. While the use of print technology remains an option, considerations of cost generally rule it out. But more to the point, Internet tools allow elaborate combinations and articulation of source documents, of text with images, of sets of images. One can weave these various elements into alternative narratives that may offer a more complete account or reach out more directly to the reader. Undoubtedly, the use of photographs comes with problems, challenges and risks. Even with the best intentions and scholarly credentials, the display of images may cause unexpected protests.¹⁹

A central issue in taking visual documents as sources to produce a narrative is whether that narrative will actually be “image-driven” or “text-driven”. By training and inclination, many historians – including myself – may tend to reproduce their textual narratives even when using images with an attempt to cover all aspects, from background to conclusion. Another shortcoming is the temptation to flow the audience with as many pictures as possible in following a logical – clinical – description of the phenomenon at stake. As I mentioned above, no single photograph and not even any extensive series of photograph will ever be able to represent more than the limited time frame when they were taken. The multiplication of pictures is less important than their degree of relevance for a given topic, after their credibility has been ascertained. A carefully selected picture – or set of pictures – may prove more meaningful for the purpose of writing a narrative or providing evidence for a point. There is in fact no reason to consider an image cannot – or should not – be used in the same way as textual documents.

It is not uncommon for a historian to retrieve information from a document that relates to a specific object at a very specific time and to either project that information onto a larger period or to insert that information into an interpretative framework that goes beyond the actual meaning of the original piece of information. Interpretation is what gives a meaning to our bits of data and imagination is often necessary to bridge the gaps left by the historical documentation. What I want to suggest here is that we should not refrain from applying this to photographs when it comes to constructing a narrative entirely or partly based on them. It may be more delicate than with textual records as the original source usually fades away and melts into the narrative produced by the historian. When using images, the image will still be there to be seen by the reader “as it was” (unless the image disappears and makes way to a commentary of the image). This creates a different configuration for the production of a historical narrative as the historian somehow loses a part of his autonomy in how that narrative will be read. Finally, while photographs can become the core source of a study, this will not free the historian from his regular practice of cross-using sources. The range between “image-driven” and “text-driven” will therefore remain very open.

The title of this paper refers to the use of photographs as a “substitute to memory”. This will sound perhaps excessive and overtly ambitious. It also means stepping into a delicate, or even dangerous ground, that of the use (and misuse or manipulation) of images to “create” a missing memory or to alter existing memories. What happened in Shanghai in the summer and fall of 1937, but also throughout the war for some segments of the population was a traumatic event. War always is. And previous experiences hardly helped as war came in new ways to different people. This experience was recounted in various texts, a few of them right after the war, most of them in official publications set in the master narrative of the patriotic war of resistance. It will be interesting to follow the track of commemorative publications after 1949 that served to build a memory of the war in Shanghai. My sense at this point is that apart from newspapers articles on anniversary dates, there was little contribution until the early 1980s when disputes arose over the revision of history textbooks in Japan.²⁰ Within that very specific framework, texts and photographs were mobilized to reveal the evils of Japanese occupation and instill a rewritten (or newly written) memory of the war among Shanghai residents (and more broadly among the Chinese people). As we all know, it was less an essay in historical inquiry than an ideological exercise in building a case about Japanese brutality. As far as the use of photographs was concerned, it followed the same pattern, leaving aside entire parts of the war experience, especially that of ordinary people.

The issue of refugees looms very large in the history of Shanghai.²¹ One could even say that Shanghai’s destiny was conditioned by the waves of refugees that sought a haven in the foreign-powers protected settlements from the mid-19^th century onward. I shall not discuss this here. My concern is with the refugees of the Sino-Japanese incidents in the city, in 1931-1932 and 1937. These were unusual refugees, especially in 1937. They differed greatly from almost all their predecessors in the previous 80 years. Actually, the word “refugees” is misleading here. Of course, this is about people who all of a sudden lost their home and found themselves thrown into the streets. Yet, these were also Shanghai residents, not outsiders, not impoverished peasants seeking refuge in the city. What we deal with here are people who lived in the very city where circumstances placed them in a desperate situation, but still a city they knew well, a city where they used to work, a city where they had connections (relatives, friends), a city where they had networks to rely on (native-place associations), a city where many were able to find ways out of their desperate situation.

This is not to brush away the extreme hardships most of these Shanghai residents experienced. The Japanese assault was a very traumatic moment. Even if the vast majority saved their life, many just lost almost anything they had. They had to rebuild themselves from scratch. Nevertheless, it is essential to factor in this aspect as it goes a long way to explain why the flooding of “refugees” in the settlements did not turn into a humanitarian disaster, why eventually only a small portion of them ended up in refugee camps, and how the complete breakdown of the city was averted. It will also help us to read and understand the images left for us to see. Observers of the time all agreed that about a million people sought refuge in the foreign settlements. It is also admitted that the native-place associations were able to evacuate about a third of this population. And last, all statistics show that the total number of refugees in camps at any one time never went beyond 120,000 people, while an estimated 75,000-100,000 lived in the streets. Even if we take into account the fact that there were successive waves of refugees (from Zhabei-Hongkou, then Nanshi, then the Western extra-settlement roads area, plus a certain amount of peasants from neighboring villages), even if we take into account the turnover in the refugee camps which would certainly push up the figure above, it becomes obvious that one way of the other, close to half a million residents-turned-refugees managed to stay in the city and survive outside of public charity support – in others words by their own means and probably with the help of relatives, friends, tongxiang, etc. The demography of the city was completely transformed as these two maps on the distribution of population before and after August 1937 reveal.

Among Shanghai residents, therefore, people took different paths on their way to become “refugees”. In fact, the place of residence itself, for instance, is part of the story. From an economic (or socio-professional) perspective, most of the people who lived in the war-stricken districts belonged to the “xiao shimin” category. It is not a very precise sociological category, but it probably encapsulates best that population. Many were also below that level as for workers, coolies, rickshaw pullers, etc. Very few were above that level. The most affluent usually lived in the foreign settlements. Those who did not had left early. We shall hardly see them in our photographs. We also know that some specific “communities” were concentrated in the northern areas, such as the Cantonese. They were probably among the best “connected” when it came to support by mutual help networks. They will show up on our photographs, though not distinguishable from the flow of refuge-seeking residents. One would not expect to find them left over on the streets. These are a few clues to read our photographs. This is part of the “blank side” I was discussed in the first section of the paper.

I shall address the issue of refugees, from the initial moment of departure amidst war and chaos to the physical extinction of specific groups, by highlighting what I try to convey as salient aspects of the refugee experience in Shanghai. This section is constructed around four sequences of photographs, each focusing on one aspect:

- Time: flight amidst chaos

- Space: segmentation and exclusion

- Space: living on the street

- Time: dying and burning

1. Time: Flight amidst chaos

War broke out on August 13, 1937 in Shanghai, but the movement of population actually started well before and, as far as Zhabei, Hongkou and even Yangshupu were concerned, was almost complete by the time fighting eventually took place. The displacement of population was not prepared or planned in any way, neither on the side of the residents, nor on the side of the authorities. There was nothing the Chinese authorities could do as almost all their resources beyond some organizational capacity focused on the war, not the residents-turned-refugees. The SMC and the French authorities had no valid reason or pretext to stop the flow as it happened before the beginning of hostilities. The population just anticipated what was about to happen. This anticipation was based on past experience, the concentration of troops in their vicinity, but it was also nurtured by groups that had an interest in such a move. While this is difficult to document, the press reported on “scare-mongers” who were going door to door to advise the local residents to move at once. There seems to have been some sort of complicity between such provocateurs, moving companies and real estate agencies. Rates as well as rents went up more than 30%. These companies took advantage of the threat of war to increase their profits.²²

We have various reports on the displacement of residents from Zhabei and Hongkou where the movement started. What the photographic record tell us is not so much how many than how and who. It also gives clues about the social background of the refugees and how they would be able to sustain the experience of the war. That the movement was massive is quite clear from that well-known picture of the Garden Bridge. It is not dated, although it does pertain to August 1937. This bridge was the main avenue through which Zhabei, Hongkou and Yangshupu residents could move into the IS. It was also the most favorable spot since the Bund and its garden offered enough space to accommodate the large influx of population. But residents also used all other bridges. They posed more difficulties due to their narrowness that caused traffic jams, even blocking all passage. It was far more dangerous for people and goods alike.

These images convey a sense of panic and chaos. People left in a rush, scared to be caught in a war situation as five years earlier. SMP statistics confirm the extent of that frenzy. Between July 26 and August 5, 50,000 were estimated to have left Zhabei.²³ On 24 August, the police counted up to 2,000 people per hour crossing along Jessfield Road on the western border.²⁴ In the French Concession, there was a more limited inflow since fighting was expected to happen in the northern districts. Yet several thousand Nanshi residents found their way into the concession before the authorities locked the gates and started to build defenses all around their territory. The NCH repeatedly used the term “hordes” about refugees, even if the newspaper expressed some sympathy toward the fate of the residents of the war-threatened districts.²⁵ These pictures testify to the massive nature of the movement of population that happened in August 1937. They can also explain why many people would lose things or even kin (children, old people) on the way.

Residents made use of all possible ways to leave the war-threatened districts. While bridges offered the most obvious passage, Soochow Creek constituted another channel, especially for those who had decided to leave Shanghai altogether. Probably, as this image shows, given the load on each boat it was very likely only the first step toward further migration back into the countryside or more simply landing upstream in the IS. The picture was taken from the southern bank of Soochow Creek. I have counted 35 people on the boat in the forefront. This is an incomplete count since the passengers in front of and on the other side of the boat are not visible, neither are those who were inside. The capacity of these boats was pushed to the limit. It is also quite clear that there is not much in terms of personal belongings, even if one can see passengers sitting on top of stuff piled up on the back of the boat. Yet the ratio between passengers and belongings is obviously low.²⁶

Most people actually left with very little. There are various ways to interpret what the pictures show to us. Residents may have thought there was only a temporary risk for their life, locked their house and left with only a few things, hoping to return to their home once the crisis was over. Some may have left in a hurry – this was probably true of most – and only gathered what they considered as valuable, or necessary, or simply moveable if they could not afford the cost of a transporter. One can also argue that many, perhaps most, actually had few items to carry with them. In all the pictures that we have, there are very few large vehicles like trucks. This picture shows that trucks were used to carry entire groups of people away, rather than one single family. The truck on this picture is filled to its maximum capacity with a man seated on one side of the windshield, holding a small suitcase. Another picture presents the same situation, although here the truck seems to be coming from a nearby village or an underdeveloped part of northern Shanghai. There is only one woman on the picture. There were few trucks available and the cost of hiring one was beyond the means of most residents. In fact, most families had few items to take with them and made use of simple modes of transportation.

The most common means of transportation for goods seem to have been hand carts or wheelbarrows and of course rickshaws. This picture provides a good sense of the density of the move, of the disorderly nature of the flight, of the incredible mix of modes of transportation, and of course of the limited “material capital” the residents possessed. It is obvious that even on a Chinese wheelbarrows or handcart one could not pile up heavy items. In fact, residents on the move gave the priority to valuables, clothing, bedding (especially bamboo mats), kitchenware and the like.

The following picture tells us as much. It is fairly representative of what a worker family would have or would be able to take along. It adds up to the bare minimum for camping, but this was not about camping. This was about surviving with small children and older people in open-sky vacant land or street alleys. Whatever the mean of transportation, people had little to move along with them. They stacked a few pieces of furniture on a small raft that took them across the river.

For the vast majority, the escape was made on foot. This picture shows a French policeman trying to monitor the flow of people. Most are walking into the Concession. By groups, by families or alone, these residents marched toward the foreign settlements with little more than they were able to carry. Even if they had more back home, what photographs show us are people with small “bundles” such as this woman with her child or the son with his old mother. The peasants came with what they were able to take on a carrying pole. There is no need to stretch the imagination to guess that these were just clothing and very basic necessities. The poor among the poor stayed behind until they were literally expelled under orders of the Japanese Navy before their assault on Yangshupu. On 8 September, 4,500 Yangshupu residents were taken by truck to the IS while another 1,500 walked their way to safety. In some cases, they had to be removed by force: “they were mostly poor people, carrying a single bundle, and apprehension was painted in the faces of most... from straw hut settlement and obscure alleyways they emerged by the hundred... there were many who had to be coaxed to get out.”²⁷

Those who possessed more than the essential had no guarantee to preserve them. The density of traffic offered opportunities for thieves or even regular rickshaw pullers when they saw an opportunity. Those who had some belongings and goods had to have them carried away on several carts or rickshaws. It was very difficult, however, to make sure that such a “convoy” would hold together. Either voluntarily or under the pressure of traffic, the pullers would split and lose track of each other, and disappear altogether, even when they had been told the final destination. More than one resident made this bitter experience.

2. Space: Segmentation and exclusion

Shanghai was not one city. It was several cities on the same piece of land. The existence of foreign settlements – two distinct jurisdictions --, that of “external road areas” to the north (“North Hongkou) and the west of the IS beyond its official boundaries, defied any attempt to make Shanghai a single urban space. Shanghai was fragmented. Probably in peacetime, this fragmentation was not too obvious. It was visible, undoubtedly, at the very least by such concrete elements such as street names, the presence Sikh or Vietnamese policemen, etc. Shanghai residents were used to live with that reality. The fragmentation of space created unequal conditions and opportunities for Shanghai residents. If we think away from our assessment of the foreign settlements as a “political space” that facilitated the emergence of all kinds of phenomena and processes we lump together as “modernization” and try to look at it from the perspective of everyday life, how then would the everyday life of a regular shimin look like? And how would it look like in a time of crisis like 1937?

Quite clearly, past experiences had taught Shanghai residents that the city was not theirs in toto or that parts of the city – namely the foreign settlements – could become enclosed spaces designed to offer protection to their own inhabitants, especially foreigners. They were equally prepared to leave out the residents of the Chinese-administered districts and even to stem any inflow of population. The natural layout of the city was favorable to the establishment of access points. On the northern side, on the northern bank, a first line of defense ran along Boundary road, with an iron railing that was reinforced in times of crisis. Only the Yangshupu district was fully indefensible, but it was not the main living quarter of foreigners (but its Chinese population was close to half a million). The strongest line of defense was Soochow creek. It separated the central districts of the IS from Chinese territory and the northern district. It was easy – and standard procedure – to block the access to bridges to prevent the population from crossing over. In the French Concession, iron gates had been installed on every street leading into Chinese territory, at least all along the former walled city. The Concession was also naturally protected by the Xuhui creek. In times of emergency, all streets on the northern bank were blocked with barbed wire, sandbags and even hard walls. Last, in 1937 (but also in 1927), the access routes on the western border of the settlements were also blocked by check points.

Spatial order had to be maintained at all times. When refugees started to pour into the International Settlement the authorities were adamant to keep some areas clean, especially in the evening. It made little sense under the circumstances, but the police had instructions to clean the Bund where refugees had congregated and to push them in to back streets. The NCH praised the SMP for its ability to enforce such instructions despite the work overload the situation had created. It even published a picture showing people who had been compelled to settle in a small street for the night.²⁸ The following morning, the Bund would be occupied again. This picture as well as some later ones also shows the difficulty to live outside: August, as we all know, is not a favorable period to stay outside in Shanghai. Temperature is high during daytime, but very often too rain comes along with typhoons. 1937 was no exception.

The central issue, however, is that Shanghainese lived in a city within which entire sections would shut down and make the inhabitants of the neighboring districts genuine outsiders. What pictures help us realize is that exclusion was embedded in the spatial fragmentation of the city. The population took its full measure in times of social unrest or war. This is the picture of a regular iron gate in the French Concession. It looks quite strong and capable of resisting the pressure of desperate crowds. Undoubtedly so, as the following images show. People are massed behind, hoping to get through or to get food when a distribution took place. The International Settlement also had its own iron gates, even if they were less systematic than in the FC, or barbed wire and checkpoints in the less urbanized Western district.²⁹

In 1937 the French Concession took exceptional measures to insulate itself from a possible invasion by refuge-seeking Chinese. Apart from the permanent gates, it actually built a wall all along its southern border in the hope to defeat attempts by Nanshi residents to jump into the concession. It did not dissuade Chinese to try their luck, as can be seen on this picture, but it definitely prevented a massive inflow. The NCH reported in details about this defense effort: “With hundreds of miles of barbed wire strung round its perimeter, supported by machine-gun emplacements and dozens of supplementary defences, the FC, true to tradition, stands ready to repulse any who dare attempt to cross the border while carrying arms. From the corner of Zikawei Road and Avenue Haig, extending to the south, the barbed wire entanglement front Zikawei creek, the banks of which are so steep and the slime so thick, that it is doubtful even the most agile could obtain a handhold....At the Route Ghisi and Avenue Dubail intersection the block houses again command all avenues while the bridge heads have been closed and triple, even quintuple quantities of barbed wire defy approach. Supporting these defences are sandbag redoubts, the machine-gun covering three directions of approach. The Concession may become a walled city... along that portion of the Zikawei creek which has been filled in, particularly in the neighborhood of the Power Plant, engineers are constructing a brick wall which faces Nantao. The construction ... is proceeding at the rate of 50 yards a day…. Further toward the business area the iron-picket gates, reinforced in front by barbed wire and behind by sandbags, are closed, ingress and egress permitted at prescribed points only”.³⁰ The FC was turned into a fortified camp. Access was strictly controlled by the police.

On all sides, the access to the foreign settlements was severely controlled, though not necessarily prohibited. Yet the more time went, the less the authorities were inclined towards admitting refuge-seeking individuals and families. For the late comers who showed up on the western borders, admission was conditioned upon bringing in sufficient food supply to maintain themselves.³¹ While the authorities of the foreign settlements definitely tried to keep a balance between their inclination to protect the population within their boundaries and opening up their gates to the residents threatened by war, they eventually enforced a policy of containment that left hundreds of thousands Shanghai residents exposed to the ravages of war. Nevertheless, the very existence of such protected areas was a magnet for the residents from neighboring districts in times of war.

3. Space: Living on the street

Probably the vast majority of the Shanghai residents who sought refuge in the International Settlement or the French Concession spent a few nights in the open air. We know that many eventually found a place thanks to friends and relatives. Evacuation was also organized, but no more than a 350,000 people actually left in late 1937 and early 1938, while the outward flow was partly compensated by new arrivals from the neighboring villages. A substantial part of the homeless were received in all kinds of shelters and camps. Yet, even at its highest, the total number of camp inmates in the IS reached 97,000 some time in December 1937.³² Of course, one needs to take into account the turnover in camps for a more accurate assessment of the total population of refugees in camps. Nevertheless, it is also obvious that a very large group of people did not find their way into camps or into proper shelters. They settled in the streets, in the back alleys (lilong), on vacant land, in any place where they eventually ended up in misery. In the IS, the SMP estimated their number to 75,000-100,000 in December 1937, but these were already only the remaining “pockets” of street refugees.³³ I do not have any figure for the FC. There must have been too, though in smaller numbers.

People set up their quarter wherever they could, including on the pavement of major thoroughfares. They moved into an area where vacant land was scarce, except in the western parts of the settlements. After they had arrived in or close to the foreign settlements, they dropped their bundles, buckets, and probably started to figure out what to do next. As mentioned above, most arrived with only the few basic items that would allow them to settle down almost anywhere, but for a short period. The following images are quite telling. People were lost. They were also often tired from a long and stressful trek, probably afraid to be turned down at the entrance of the settlements, and simply traumatized to have left or lost their home, job, etc. Perhaps, people were tired after days and nights spent in such conditions. It is delicate to interpret such emotions from photographs. There is a risk of overinterpretation, but my guess is that many if not most must have felt a sense of loss. There was little they could do, especially if they belonged to the less organized groups in the city, the groups that could not turn to the more or less powerful native-place associations to help them through their ordeal.

It is impossible to make any interpretation of this on the basis of photographs (we also have pictures of refugees being evacuated in proper order by their native place association). Yet these images show people who had nowhere to go, who set up temporary quarters in a back street where they could afford only very rudimentary comfort and a relative protection from traffic, public view or simply the weather. But sometimes, they just stayed where they were, on the pavement of a public thoroughfare, for all to see, oblivious to the ongoing activity around them.

Many images show babies and young children. They are almost everywhere on our pictures. Written sources do not account accurately for births in Shanghai. In fact, as I argue below, many simply did not live long enough to be recorded in any way. Children were the most vulnerable ones among the refugee population. Living on the street with little or no resources left them with little hope of survival despite all efforts by their parents. This picture of a young mother feeding her baby is just one case of a woman who is quite evidently in a state of complete poverty herself and may not be able to breast-feed her child. This single picture, however, gives us a measure of the immense misery and suffering the less privileged residents of Shanghai went though with their forced departure from home. Although high infant mortality was a general phenomenon in urban areas in the Republican era, war had devastating effects on the children of these poorer classes.

4. Time: Dying and cremating

War was a time of death. Death for those who were involved in military combat. Death for civilians caught in the middle of bombing, shelling and fire. But also death for the displaced people made more vulnerable to disease, even benign illnesses that eventually killed weakened and undernourished bodies. This aspect is hardly documented in public documents like newspapers or reports by charity organizations. It can be retrieved by the historian digging up statistics from forgotten archives. These figures, however, may not convey the full extent of the tragedy that was taking place in Shanghai at any time, but which the war situation pushed to unbelievable heights. In this section, I present a set of disjointed pictures and, on the basis of this, make several assumptions that, I believe, do reflect what actually happened in Shanghai in 1937-1938.

My focus is on the fate of the refugees that failed to find a proper place to survive the time of hostilities. Of course, even among those who managed to get into camps, death struck steadily at higher rates than usual. It is also true that even in peacetime, the poorest groups in the population would just abandon their dead – adults and children – on vacant land, on the street, almost anywhere, in flimsy coffins, bamboo matting or just their clothes. In other words, there was an ongoing process of people, mostly infants, dying in the streets and dumped in the streets of Shanghai at any one time in the late imperial and republican period.³⁴ During wartime, however, the number of such deaths was multiplied several times and I argue that most likely these deaths occurred among the refugees left on the streets. I shall try to document this through images at the same time as I shall draw my information from these images.

There are, of course, few pictures of abandoned bodies. Only the authorities made it their policy to take snapshots of dead adults for purpose of identification. They also checked whether it was death from natural causes or from an act of violence. These records left us with thousands of ID pictures of little use for the historian. Newspapers generally skipped the topic of abandoned corpses altogether unless it became an annoying problem for residents, especially foreign residents. Photographs were less likely to be published. In other words, these invisible dead would rarely show in print. This image published by the North China Daily News is that of the body of an abandoned baby that remained unpicked for several days right at the entrance of a major building in the IS. The quality of the image is very average, but one can see the head of the baby, the type of matting in which small children were usually wrapped, and the location where the body was left.

Such abandoned corpses or coffins – corpses or coffins “without master” (無主屍體 in administrative parlance) – were counted by the thousands, then by the tens of thousands. They were collected by two organizations, the Shanghai Public Benevolent Cemetery (Pushan shanzhuang – 普善山莊), mostly in the IS, and the Tongren fuyuantang (同仁輔元堂), mainly in the French Concession. The Shanghai Public Benevolent Cemetery was by far the leading operation in Shanghai.³⁵ They maintained a staff of “body collectors” that patrolled the streets of Shanghai from dawn to night to pick up bodies in areas known to be deposing spots, or under instructions from the authorities, or following a resident’s call. They also picked up the dead without resources or relatives in hospitals, other charity organizations, etc. Their main activity, however, was securing corpses left in the open air in the streets of Shanghai. They used various tools, but the most common one was handcart such as the one shown on the picture. The scene is taking place in the French Concession in daytime. The body collectors are easily recognizable by their “uniform” with 同仁輔元堂 printed on their back. On this picture, given the effort the collectors seem to be making, it must be an adult. Nevertheless, all statistics show that the vast majority of abandoned corpses – 85% on average – were those of infants and small children.

This picture of a collector of the Tongren fuyuantang was not taken in 1937-1938, but in 1947. I use it because it shows a reality that makes no difference with wartime. The only obvious difference is the use of a pedicab instead of a handcart. Pedicabs were introduced in the early 1940 in Shanghai. The photographer, Jack Birns, misread what he was taking. He labeled his photograph “A bicycle cart delivers a child's corpse to a temporary morgue”.³⁶ He missed the fact that this was a collector from the Tongren fuyuantang doing his regular job of picking up bodies and delivering them to one of the organization’s stations for encoffining. This was not a “temporary morgue”. Despite the temporal shift, this image, I argue, does help us seeing and knowing what it was like to live in Shanghai 1937-1938 when such bodies were collected by the tens of thousands. The second major duty of the Tongren fuyuantang was to provide coffins and perform the appropriate rites before burial. As can be seen on the following image, considerations of cost – as true in the civil war period when Shanghai received a new wave of refugees as during the Sino-Japanese war -- caused the Tongren fuyuantang to put several infant bodies together in one coffin. Again, the original caption is misleading: “Children's corpses in a collective coffin await cremation on Christmas Eve”.³⁷ I doubt, at this point, that the bodies of indigents were cremated in the late 1940s.

Once the coffins were ready, they were transported to their place of burial. In normal times, the bodies collected by the Tongren fuyuantang were buried in one of its 13 cemeteries located in Pudong. With the war, however, these areas remained out of reach until the end of 1938. While the authorities provided vacant land in the western outskirts of the settlements in Chinese territory, the number of people dying in the streets went beyond what these plots of land could accommodate. Faced with the issue of having to “host” the coffins of all the dead who could afford the cost of storage in a private or a guild repository – eventually more than 100,000 were stored in the International Settlement– the authorities of both settlements decided to impose the cremation of all abandoned corpses, the protest of the charity organizations notwithstanding. Coffins were transported to two or three different places in Western Shanghai where they were piled up on a stack of wood and set afire with gasoline. This was performed under the surveillance of official representatives of the SMC or FMC. It was exceptional that the head of public health division (here Jean Malval) would be present himself as is the case.

The picture tells us these are the coffins of indigent people. It also confirms, by their size, that most contained the bodies of infants and small children. Adult coffins are few and stocked underneath the smaller coffins. The process took a few hours as various stacks of coffins were piled side by side and left to burn until there were only ashes left. This is a rare instance of photodocumenting an otherwise quite and almost invisible process.

Concluding remarks

This paper has tried to address the issue of using photographs both as a source and as a medium to build a historical narrative. While it is driven both by text and images, I believe photographs are a significant part of its construction. There would be no point in excluding elements that can inform us on the issue of refugees and to pretend to elaborate an “images-only” kind of narrative. What the Internet offers is precisely the possibility to combine various types of documents in a text, but also to juxtapose different forms of narrative. This paper incorporates individual images as well as individual maps that serve to provide visual support to the arguments made in the text. This is not just illustration as my exploration start from the images, especially photographs, themselves. I have been drawing on their content to understand how the process of becoming homeless unfolded in the particular context of Shanghai and what some of the consequences of this displacement were. I also hope to have shown that photographs can take us closer to aspects of life that other forms of record may have overlooked or missed entirely.

The experience of war was not just an accidental event for some areas of Shanghai. It was in fact an integral part of living in districts such as Zhabei or Hongkou. The end of the Sino-Japanese war did not mean the end of such migration within and outside of the city. While current official history celebrates the liberation of Shanghai in 1949 and the warm welcome the PLA received, independent photographic records have documented the departure en masse and by any means of the local population. These images are reminiscent of those presented in this paper. War is a terrible threat many people would simply try to avoid by fleeing to other places. 1937 certainly represents the most momentous episode in this experience. The photographs included here display how much chaos, uncertainty, and suffering the war brought about. They also materialize quite vividly the social inequalities associated to the political divisions within the city. The traditionally densely inhabited districts at the periphery of the foreign settlements were home to a population with limited resources who could hardly survive without outside help. If they happened to fall out of the protective net of kinship or native-place networks, their future was bleak and that of their children bound to be short.

Fighting did not last very long (three months) but it had devastating effects for years, either on the living and economic conditions of the population or on the physical makeup of entire areas of the city. Photographs highlight not just the most gruesome aspects of war – actually there was little of this except fro the accidental bombings in the foreign settlements – but also the “everyday life” experience and challenges common people faced, how they dealt with their difficulties, what they “chose” to do to survive, what they were able to save or most concerned about. Most of these pictures are not images of heroism. These are utterly powerless people stuck by a bad fate. And yet these photographs are a testimony of the courage and endurance of ordinary people caught in a vicious war. As much and as fast as they could they fled to save their lives, and whatever little belongings they had. Many perished, especially among overexposed children. But most were able to reestablish themselves, either in their native places or even in the city proper. It was a subdued and hardly perceptible victory over the enemy no one ever claimed. Its memory is now gone. Yet for these “local refugees” it was a daily fight to keep themselves and their families alive. Our photographs do not document that process, or hardly, but they help preserving a concrete trace of a forgotten past.