1 Nothing can be more frustrating or more rewarding to the researcher in material history than archival research. Canadian archives contain a wealth of primary and secondary sources which reveal to the diligent an immensely detailed picture of past mores, manners, and practice. Much of this wealth is as yet untapped; researchers are few and time scarce. Yet this exploration of the written documentation of the past is vital if we are to attempt to understand the world as our ancestors apprehended it.

2 In Newfoundland the case is particularly acute. The last general history of the province was written in the 1890s. Since then historians have for the most part concentrated on the unravelling of the province's complicated political saga, its economic position in Atlantic trade, or the development of its institutions. Only recently have certain geographers, folklorists, and material historians begun to research the social and cultural patterns of Newfoundland and Labrador. While studies in cultural geography or folklore are rare, published research in material history is almost non-existent. Shane O'Dea has compiled lists of native cabinetmakers and silversmiths; John Joy has examined the later nineteenth-century manufacturing base in St. John's in an unpublished M.A. thesis for Memorial University; and John Mannion's students have begun to look into the cloth trade and clothing. Perhaps the most extensive studies with a bearing on material history have been conducted by the Maritime History Group under Keith Matthews. Researchers have painstakingly transcribed thousands of shipping lists in order to build a picture of the Newfoundland trade.

3 When the Newfoundland Museum came to plan its new exhibits, however, staff discovered that little documentation existed on the collections themselves and secondary sources on material culture in Newfoundland were unavailable. Thus it was necessary to turn to primary archival documents for basic information on provenance, type, and date of introduction of goods imported or variety, type, and maker of goods produced in Newfoundland. John Joy had examined the directories and his thesis on manufacturing in St. John's was most helpful, but the directories began in the 1880s. Matthews's and Mannion's import lists were excellent but they were not as specific as the museum could hope. Probates were available, as were business ledgers, and the Geography Department had begun work on these. The museum decided to turn to the advertisements in Newfoundland newspapers to extract information primarily on commodities and services available in St. John's and other outport communities throughout the nineteenth century. As staff began to examine these sources, it became obvious that they were covering ground covered often before. Students had looked through the same tattered copies of the Public Ledger or the Carbonear Herald for information on the courts, on temperance meetings, on reports of scandal. Shane O'Dea had scanned the journals for native cabinetmakers and practising silversmiths. It became clear that after we had finished our work, yet another researcher would sift through the same material, looking for a slightly different set of data. Was there a simple mechanism by which much valuable research time could be saved and needless duplication of research efforts be eliminated? With this goal in mind we began to look closely at the computer services of the National Inventory Programme, National Museums of Canada, and the possibility that its programmes and our needs might indeed mesh.

4 To many trained in the humanities the use of a computer in conducting research is anathema. There is a fear that the use of a machine will distance the researcher from the material, making the creative interchange that so frequently happens between the document and the historian impossible. The computer is, however, an accepted research tool in the social sciences and in archaeology. While it can never replace well-trained and curious human intelligence, it can act as a valuable tool for that intelligence, performing the menial and time-consuming research tasks that senior researchers normally delegate to graduate students and assistants.

5 It is important to understand the nature of the computer's function. The English word "computer" implies a machine that figures or thinks; the French term ordonnateur is perhaps more appropriate. A computer at its most basic is an orderer, an organizer of information. Its great advantage over the human organizer is its ability to digest immense quantities of disparate data and to order them in a seemingly endless variety of ways. This ability has obvious practical applications for the problems of archival research. If information from an archival source can be entered into a system where it can be indexed in numerous ways, the same body of material can serve more than one user.

6 This can be illustrated by returning to the use of newspaper advertisements in material history research. Different researchers will examine these for different purposes. One researcher may be interested in the variety of foodstuffs, both imported and locally produced, available in St. John's in 1814. Another researcher may be interested in the names of the local merchants who retailed food produce. Another may be concerned with the variation in foodstuffs imported over a thirty-year period from 1814 to 1844. All three researchers will use the data gathered to answer totally different questions. By placing data from these same advertisements into a standard computer format, it is possible to ask the computer to search the data set, retrieve the data relevant to each question, and organize it for the particular user. Once the data has been entered into the computer, the operations of retrieval and organization can be accomplished in a matter of seconds. Thus, instead of three graduate students wearily examining the same rolls of microfilm or yellowed, crumbling pages and producing three specific reports (none of which can be used to answer any questions other than those originally stipulated by the researcher), the same data base, entered once, can be indexed by the computer to yield not three, but countless answers to specific problems.

7 Researchers are, however, notoriously idiosyncratic. Could a single data base, properly indexed, really fulfill the needs of more than one researcher? It is important here to realize that, unlike the research report produced to answer a particular query, material entered into a computer is not entered according to specific end-use. Once a source has been decided, all data from that source are entered indiscriminately. No judgments are made at the point of entry; thus, all the material in the original source is available in the computer's memory banks. The researcher can then call up any of the information entered and in a variety of configurations. Returning to our example of the three researchers interested in various aspects of food consumption and marketing in nineteenth-century St. John's, we can illustrate this recall ability.

8 Below are two examples of typical advertisements (see Appendix 1 for copies of advertisements in computer print-out form):

9 Researcher A, interested in the variety of foodstuffs sold in 1814, could ask the computer to search out all foodstuffs listed in such advertisements throughout the year and would receive a printout listing "prime corned beef, prime fresh beef, mutton, cider, potatoes, turnips, onions, oats, turkies, geese" and any additional items advertised that year. Researcher B, interested in local merchants selling food in 1814, would request a search and receive a list with the names of "Parker, Bully, Job and Company," etc. Researcher C, attempting to compare the variety of items imported over a thirty-year period, might ask the computer to print out all food items advertised every January and every July, every other year, beginning with 1814 and ending in 1844.

10 These are very simple retrievals. A research problem can be extremely complex. Researcher A may in fact want to compare what was sold in Carbonear with what was sold in St. John's and may also want to make a seasonal comparison over a five-year period. Researcher B may require not only a listing of merchants who sold local produce, but also a cross-index of specific merchants with specific kinds of produce. Researcher C may demand not simply lists of items but also a correlation with ports of origin and merchants over the thirty-year period. Admittedly none of these problems would be beyond the competence of three good researchers, but none could perform his searches as quickly as the computer.

11 Initial discussions with interested groups in St. John's, notably the Maritime History Group at Memorial and Parks Canada, indicated that a computer-based research format for advertisements could prove useful. The National Inventory was willing to explore the problem with the museum and to aid in the development of an appropriate format. The computer's strength is in its speed, efficiency, and versatility in sorting information; its limitation is its inability to go beyond the given. A computer can only retrieve what it has received. In order to retrieve data, it must receive information in a consistent manner. A computer cannot analyze idiosyncratic input. Thus, if a computer is to be asked to compare data (for example, the items sold in St. John's and the items sold in Carbonear), it must receive that data in the same manner.

12 The development of the format for recording information is obviously of prime importance in developing a workable system. A format must answer several criteria. It must be machine readable: in other words it must record information in discrete units, or fields, that can be indexed by the computer. Secondly, it must divide the information into as many fields as possible so that it can allow the computer to search a large number of fields and thus provide responses to as many different queries as possible. Thirdly, the format must be relatively simple to complete. If information must be encoded into a mathematical formula in order to be entered into the computer, the time-saving advantages of computer indexing are lost. Moreover, the information entered in code can only be retrieved in code and the process of decoding is also time-consuming.

13 The last criterion was easily satisfied. Canadians are fortunate that the National Inventory in Ottawa uses a computer programme which answers the need for simplicity of coding and retrieval of information. The ISIS programme is a word-based system of data entry. The Inventory uses this system to record information on artifacts and specimens held in various museum collections. One aspect of this programme, applied primarily to the cataloguing of information on works of art, was admirably suited to the recording of information in newspaper advertisements. In order to record subject matter in paintings, a syntactical system was developed. For example, in cataloguing a landscape painting the cataloguer chooses the primary subject matter, "waterfall," and that word becomes the prime term. This is the term which describes in general the content of the painting. This is particularized by the addition of descriptive phrases after the prime term, separated from it by commas, the whole description ending with a period. The computer "reads" the first word entered as the prime term, understands the words following the comma as secondary descriptive terms, and recognizes the period as the end of the description.

14 This method of recording works admirably with advertisements. Take, for example, an advertisement for 100 pounds of the finest quality lemons appearing in a St. John's newspaper of 1850. The researcher records "lemons, 100 pounds, finest quality." Anyone looking for the occurrence of lemons can simply search for the prime term "lemons." There is no necessity to code for lemons or to add informative terms as in "fruit; lemons."

15 Once it had been ascertained that this descriptive system would record information from advertisements without enormous difficulties, it was then necessary to decide on the kind of information to extract and the number of fields into which this information would be placed. Though the interest of the museum was in the commodities themselves and the date of their appearance, it was recognized that other users might be interested in different types of information appearing in the advertisements, such as the name of the importing merchant, the vessel which carried the goods, or the appearance of an illustration in the text. Trial and error produced a draft format of twenty-five fields. An initial test run of 100 advertisements produced some modifications in the original design, but the present revised format (see Appendix 2) has worked very well.

16 From the information recorded one can learn the name of a merchant, the location of his business, the type of his business (for example, commission merchant, merchant-retailer, auctioneer), the vessel on which his goods arrived, the provenance of the shipment, whether or not he was advertising a new shipment, a fire sale, or a spring clearance, and the quantity, quality, and very often the precise colour and style of what he sold. Similarly the format can record the items offered for sale at an auction or the services advertised by a local craftsperson. Interesting tidbits of information, such as the fact that fish or oil can be exchanged for goods or that the merchant caters specifically to the outport trade, can be entered in field 22 under "Remarks." It must be emphasized that there is no restriction on searching; the format can be searched on all twenty-five fields. For example, if a researcher were interested in a particular vessel, the Cluthia, he could request a search on field 9 ("Carrier") for the incidence of the Cluthia as a carrier. (The museum, which has several watercolours of this vessel, is requesting such a search. Together with enlarged photographs of the advertisements located by the computer and a copy of shipping lists, the paintings will appear in an exhibit on the Newfoundland trade.) The chief problem to be overcome in format design dealt with what the museum considered to be the most interesting information - commodities and services. The resolution of this problem has enhanced the usefulness of the data stored. Even before the test run it became obvious that a major difficulty was the initial categorization of the seemingly endless commodities listed in the advertisements. To have listed all commodities as prime terms under a single field ("Commodity") would have required the researcher interested in foodstuffs to search for an endless number of specific items. It would have been impossible to request a listing of all food items sold in one month, since it was unlikely that the researcher could have pinpointed precisely the items sold.

17 This problem resolved itself very neatly around a suggestion made by David Alexander of the Maritime History Group. The format utilized the broad categories compiled by Statistics Canada in its "Trade of Canada Commodity Classification." In order to analyze present import-export patterns, Statistics Canada has devised a number of general categories under which an unbelievably disparate selection of goods may be subsumed. Museum and National Inventory staff modified this system slightly to reflect nineteenth-century goods and developed the following nine broad categories for commodity classification:

We also added a "Services" classification to record the announcements of photographers, music teachers, hairdressers, cabinetmakers, and so on. The Statistics Canada classifications have proven very satisfactory and though specific goods differ it is possible to refer to the Statistics Canada expanded lists if in doubt as to the very general category in which to place an item. Since the formats are relatively consistent, this solution has also made possible the comparison of nineteenth-century imports with contemporary imports through computer indexing.

18 With National Inventory assistance the museum was able both to test and to use the format as a research tool. Deborah Jewett of the Inventory analyzed a test run of 100 advertisements and assisted in the revision of the format; she also co-ordinated the development of ancillary materials - "Any Lists," glossaries, instructions - necessary to the actual project. The museum was fortunate in hiring, with the Inventory's help, a Canada Works summer team. This team was led by Valerie Kolonel, a history graduate from Memorial University, who managed four researchers and a data entry operator over a four-month period, May - September 1979.

19 As the first newspaper in Newfoundland was published in 1806, it was decided to record advertisements appearing from that date to 1900. This required the analysis of some 43 newspapers from St. John's, Carbonear, Harbour Grace, Heart's Content, and Trinity. Most were available on microfilm or in bound volumes at the Provincial Archives or Memorial University. During the summer the researchers recorded between 40 and 60 advertisements per day, completing 6,979 formats in all, while the data entry operator entered 2,625 of these into the computer. (Data entry is continuing at the museum.) The newspapers and the years covered are listed in Appendix 3 and a completed format is shown in Appendix 2.

20 In addition to managing the group the co-ordinator was also responsible for the compilation of glossaries and, with the help of the Inventory, of Any Lists. It became apparent as work progressed that descriptive terminology for goods and some services had changed radically since the nineteenth century. To ensure that an item was recorded in the correct field, the co-ordinator was obliged to develop extensive glossaries which researchers might consult. For example, what were "Norway rags?" Should they be placed in field 15 ("Fabricated Materials") or field 19 ("Personal and Household goods")? (They are cutstones for building.) Abbreviations for items or weights and measures were often obscure. These too are included in the glossaries, along with their contemporary equivalents.

21 One of the most important aids for the user are the Any Lists. Almost sixty such lists were developed to group similar commodities under broad headings. For example, one of the Any Lists for field 13 ("Food, Feed, Beverages, Tobacco") is "Beverages." This list includes the prime terms entered on the format, for example, rum, muscat, muscatel, vin-de-grave, cider. Thus, a researcher interested in any beverage advertised in 1831 in St. John's can request the computer to search field 13 for "Any Beverage" and receive a printout listing all the advertisements containing any one of the specific beverages included on the Any Lists. The list also allows the researcher to make a quick check to see if the item in which he is interested has ever been recorded. If "cider" does not occur as a prime term on the Any List for beverages, then no incidence of cider has been recorded in an advertisement. Any Lists also include alternate spellings, if they have been recorded as prime terms, for example, "carraway," "carroway." The flexibility of the Any List eliminates the need for the researcher to decide on the correct spelling of a prime term; the item can be recorded exactly as it appears in the source. It should be noted, however, that if all occurrences of an item have been recorded in one spelling, for example, "trowsers," the contemporary researcher who requests a search for records of "trousers" will be disappointed. The computer cannot recognize the difference in spelling although it will compensate for incidents of plural and singular records. An Any List for beverages plus a table of all available Any Lists appears as Appendix 4.

22 Material entered into the computer can be retrieved in a variety of ways, the simplest being a printout of the records. The Newfoundland Museum has printouts of all records entered. These and the Any Lists have already been used by a number of researchers with simple requests such as "Do you have any records of advertisements for aerated water?" More complicated requests for cross-indices, lists, and so on are best answered through a direct request to the computer. If, for example, a researcher wanted to know the types of beverages available in 1831, he would request a search on field 4 for "1831" and on field 19 for "Any Beverages." The computer would then produce a printout of the advertisements for beverages in 1831. The actual mechanical process is slightly more complex, since it involves the terminal operator's ability to phrase the question correctly to the computer. But the printout itself is verbal and immediately accessible to the researcher. The researcher familiar with a terminal can in fact request the information to appear on the terminal screen, make the requisite notes, and then request a printout of the particular items desired. The retrieval process can also be very specific. If a researcher is interested in à particular item, such as Belgian carpets, he can request a search in field 19 for "Carpets, Belgian" and receive any instances of this item.

23 A researcher using this system has available literally at his fingertips a vast body of information, retrievable almost instantly. The information can be cross-indexed automatically to correlate particular merchants with particular goods, vessels with merchants, provenance with commodity. Items can be grouped by year, lists of particular items generated, craftspeople enumerated. It would be a mistake, however, to view the information thus retrieved as providing the ultimate authority on the subject. If a search reveals no Belgian carpets in 1831, that does not necessarily mean that no Belgian carpets graced the salons of St. John's in that year; it just means that they were not advertised. A computer-based system such as this is simply a tool, useful to the researcher. Information from this system combined with that from ledgers, probates, diary and journal accounts, and the artifacts themselves, can help to build up a more complete picture of the past. The computer cannot substitute for careful research, only assist it.

24 With the co-operation of the National Inventory Programme and the sponsorship of Memorial University, Valerie Kolonel and museum staff are presently experimenting with the system and compiling a user's guide which is expected to be available this summer. Although the system is operable, it is at present limited by an incomplete data set and further work will be required to analyze all nineteenth-century newspaper advertisements. (It may seem fruitless to record every advertisement for 100 years, but it should be noted that the only reference to the work of the daguerrotypist, A. Salteri of Halifax, in Newfoundland, occurs in one entry in 1852 in a Carbonear paper. Prior to the location of this entry, his presence could be inferred, but not proved.) Possibilities also exist for expansion of the computer-based format to record ledgers and probates as well as for the development of programmes to compare computerized shipping list data, as recorded by the Maritime History Group, with computerized advertisements.

25 In the final analysis it must be determined whether the effort required for recording and entry of information is worthwhile. This will only be truly proven through further experimentation with this preliminary study. Already, however, users of this first rough system have given encouraging reports. The use of the system is not limited strictly to research papers, but has applications in the exhibition and restoration field. The museum uses the information to locate appropriate graphic material to complement artifact exhibits. Parks Canada and the Provincial Historic Sites branch see the system operating as a "shopping list" to aid in determining period furnishings. Similar studies in other Atlantic provinces may eventually provide a computer-indexed picture of food habits, styles of dress, and patterns of service provisions. The Newfoundland Museum and Memorial University welcome any enquiries or comments concerning the project and will be delighted to send further information.

26 Please address all enquiries as follows: