1 During the academic year 1983-84, a graduate history seminar at the University of New Brunswick in Fredericton attempted to construct a methodology for the analysis of artifacts.1 This was the first seminar in what would become the Diploma Programme in Material History offered in conjunction with the M.A. Programme in History at the university.2 The course was taught by Dr. Stuart Smith of the Department of History whose creative leadership provided the climate for group discovery which has resulted in this report.

2 The article has been divided into three sections: a description of the class process in developing a methodology and research model; the research model described; and a test case application of the methodology using a sample artifact. No attempt has been made to integrate this line of inquiry within the broader questions posed by the discipline of history. Rather, it is presented as an end in itself and, as such, will hopefully serve as a useful catalyst for discussion and a departure point for those interested in material history and the analysis of artifacts.

3 The search for a workable methodology began with a general, surface investigation of several past cultures and attitudes held towards the object in those cultures. This process was repeated, more or less, when a number of social science disciplines were examined and their relationship with material evidence measured. Of this group, archaeology appeared the closest methodological link to the class objective in view of its basic procedure of commencing investigation with the object, a direction no doubt born of necessity because of the paucity of other forms of evidence available in the course of the archaeologists' work. In spite of or perhaps because of this factor, the seminar adopted archaeology's initial stage of scientific description of material evidence as a core philosophy.

4 Equipped with this outlook the class then began to digest and qualify a number of methodological proposals contained in pertinent literature that attempted to formulate a line of enquiry for the analysis of artifacts. Two of the more promising models appeared in recent issues of the Winterthur Portfolio - E. McClung Fleming's "Artifact Study: A Proposed Model" and Jules Prown's "Mind in Matter: An Introduction to Material Culture Theory and Method".3 Aspects of Fleming's methodological framework were in fact altered and adapted in a generalized sense for use in a preliminary class model emphasizing several basic properties of the object. Fleming's basic artifact properties included History, Material, Construction, Design and Function. To each of these he applied four analytical operations: Identification (factual description), Evaluation (judgements), Cultural Analysis (relationship of the artifact to its culture) and Interpretation (significance).

5 Upon reflection, the class decided to reduce the perceived complexity of this framework by abandoning the four analytical operations and creating a cleaner, basic model with aspects of Fleming's methodology intact. History, for example, was omitted as the initial artifact property and replaced by Material, a move that seemed consistent with the archaeologist's method of beginning analysis with the artifact. Next, as in Fleming's model, Construction (including physical description) would be considered followed by Provenance (History). Design was eliminated in the belief it would be incorporated by Construction and other elements of the model and therefore did not warrant separate classification. Function was viewed as essential to determine the artifact's use and what implications, if any, were intended or unintended through that use. Finally, a property termed Value was added to the basic framework. This was seen as the most interpretive portion of the model despite its shallow monetary connotation. Aside from the price an object might bring in a contemporary auction setting or the consideration of its purely aesthetic value, this property represented a more complex level of analysis and was related, in large part, to the object's cultural associations with, and perceived value to, the society in which it was produced. At the same time, it was recognized that an artifact's value could be interpreted differently by a range of observers from its point of creation to the present day.

6 Material, Construction, Provenance, Function and Value thus formed the seminar's core model. These properties would be considered as listed, an order that anticipated the examination of an artifact as the starting point of analysis before the consultation of supplementary source material. The arrangement of the properties also reflected a gradual shift from the more empirical observations gained in Material and Construction to the largely interpretive property of Value.

7 The seminar viewed the basic model structure as a necessary retreat from Fleming's more complex proposal, which it felt attempted to synthesize too much information from several sources too early in the investigation procedure. Class members did note that their model might itself develop further through discussion and testing but anticipated a slower, more controlled evolution to the interpretive aspects of artifact analysis.

8 A more central question confronting the seminar at this point, however, was the model's adaptability. Could a broad range of artifacts be approached using this line of enquiry or was it necessary to develop separate methodologies for different categories of artifacts? Indeed, was it even necessary to categorize artifacts for analytical purposes? Jules Prown, in his article, felt that artifact categories were useful because of the broad range of material produced or modified by man. He based his classification system on function and listed several categories which progressed from the decorative to the utilitarian (i.e. Art, Diversions, Adornment, Modifications of the Landscape, Applied Arts and Devices).4 Proceeding from this arrangement, he proposed a methodology in which direct contact with the artifact would be established using the analytical stages of Description, Deduction and Speculation followed by a programme of research designed to validate these stages by considering external forms of evidence. That evidence would be gathered through a variety of methodologies and techniques developed by established disciplines (i.e. social history, cultural geography, social anthropology, etc.).5

9 Prown emphasized that these methodologies should not be applied until the artifact itself was thoroughly analyzed, a direction supported by the seminar. At the same time, however, it was noted that Prown's model could be tightened considerably, particularly the Speculation stage, which involved the formulation of theories and hypotheses based primarily on physical evidence. The seminar viewed these processes as having application near the end of the investigation and felt that Prown proposed their introduction too early with less effect. If anything, Prown relied too much on physical evidence, posing questions it could not answer. In addition, his system of artifact classification appeared, without testing, to be redundant and potentially harmful to the investigation process. Once a group of artifacts were categorized, for instance, they might receive questions not asked of artifacts in other categories and vice versa. This perceived diversity was thought dangerous to the seminar's desire for a flexible research model. It was instead decided to test the class approach on a sample artifact chosen by each student for analysis, an exercise that would hopefully serve to indicate if the model required any alterations.

10 The initial test produced some interesting results, as well as several problems, and ultimately, the need for an expanded model format. The examination of the assigned properties of each artifact produced a number of instances in which the hands on approach applied in Material and Construction yielded observations contradicting held assumptions. For example, a bowler hat's material composition, component parts and their assembly, revealed moulded cardboard as a base material with a circle of wire beneath the brim to help maintain the hat's shape. The seminar member who examined the hat had always assumed that bowlers were made primarily of felt, but these new findings contributed directly to subsequent observations regarding the hat's perceived value to the society in which it was produced.

11 Next, during consideration of Provenance as an artifact property, a British army sword was presented and the question posed as to whether it had ever been used in a major historical event such as the War of 1812. In discussing this matter it was eventually agreed that the sword's individual history and connoisseurship were less important than its more general connections to the era in which it was produced. If swords like the example presented were used during the War of 1812, then study of that artifact would likely reveal certain conclusions about the general nature of warfare at that time. The sword's property of Function, beyond its implied basic use, would consider the weight of the weapon and its place in a soldier's pack. Did it in fact retard movement? Was it awkward to swing in use, etc.?

12 The properties of Function and Value were confirmed as being more productive in an interpretive sense than expected. Value, in effect, was seen as the collection point for the buildup of information through the preceding stages. Hence, for example, elements of the bowler hat's construction technique contributed to observations regarding the values of the society that produced and used this form of head wear (uniformity, durability, formality, etc.).

13 While seminar members felt that the model performed reasonably well during this run-through, a number of problem areas were discovered. The gradual buildup of information through the artifact properties, although seen as essential, also created uncertainty in some cases as to where one property ended and the other commenced. Function seemed to flow into Value and elements of Provenance became confused with Function. In addition, the exact time and place for the introduction of supplementary data remained unclear. This included comparative data derived from objects similar to the artifact under consideration as well as documentary and other evidence in support of (or contrary to) information revealed to the observer through the hands on approach encouraged during Material and Construction. In the course of the presentations, seminar members usually began introducing supplementary data during Construction after having made initial observations derived from a direct examination of the artifact. A mixture of observable and supplementary evidence then continued through the remainder of the properties from Provenance to Value.

14 Finally, an area of major concern developed regarding the perception that some artifacts yielded more information than others when treated by the model. A number of seminar members subsequently proposed that time could thus be spent more profitably analyzing artifacts of higher information value. Others, however, felt this point of view seemed reminiscent, in some respects, of Prown's artifact categories and their progression from the decorative (aesthetic) to the primarily utilitarian. To Prown, ultimately, the decorative/aesthetic dimension of objects as embodied primarily in fine art, architecture and the applied arts resulted in greater communication value between the observer and the original producing culture than the more utilitarian artifacts, which he referred to as devices. 6 As an art historian it was felt, perhaps unfairly, that Prown had advanced his opinion because of his professional background. Yet the core of his argument confirmed the seminar's observation that the analysis of some artifacts was more profitable than others, whether this conformed to Prown's aesthetic-utilitarian construct or not. After much discussion, it was ultimately decided to abandon a scale of artifact information value since questions raised through the application of the mode were seen as equally valid whether applied to a mid-nineteenth-century hammer or a late eighteenth-century landscape painting. The landscape might yield more information about the culture that produced it, but if the hammer was not questioned at all, potentially useful information might be lost. In addition, any test of the model's flexibility would be skewed in one direction.

15 Although this issue was resolved, earlier concerns regarding the lack of a clear division among the model's artifact properties and the place of introduction of supplementary data seemed to indicate a need for structural change. A partial answer to these difficulties came during one of the final class presentations in which a nineteenth-century firearm was analyzed by applying a series of three general question categories to each artifact property in turn, from Material to Value. These categories each contained very general standardized questions in view of the broad range of artifacts the model would conceivably seek to analyze. For the sake of convenience, these were labelled A, B and C, where questions marked A involved those that could be answered through direct observation of the artifact, B those that could be answered through-comparisons with similar artifacts, and C when supplementary evidence such as printed or written sources were consulted. In addition, a final non-question category, labelled D, was created for the formulation of conclusions derived from the preceding questions.

16 This standardized format appeared to give each artifact property more definition through the application of an ordered series of question categories. Introduced in turn to Material, Construction, Provenance, Function and Value, the A, B and C questions also served as a guide for the timing and use of observable, comparative and supplementary data. Given the fact that most seminar presentations to that point had attempted to answer unordered questions of no particular type, the most recent approach was seen as a promising direction.

17 As the autumn term ended, further development occurred when it was proposed that the model might achieve greater clarity if represented in graphic format. This was suggested in the shape of a grid system which would contain the five artifact properties along its vertical dimension and the three question categories and conclusions along the horizontal dimension (see Table 1).

18 When the next term commenced, the seminar continued its examination of the proposed model. After considerable reflection and discussion it was eventually realized that the methodology, though heading in the right direction, exhibited problems and required additional modification. The existing procedure of examining an artifact through each of its properties in succession meant, for instance, that material composition was subjected to three groups of question categories and a set of conclusions derived therefrom before the property of construction was considered, and so on. In effect, the model did allow for the accumulation of knowledge about an artifact but did so at a tedious rate with a repetitive, rigid manner.

19 A solution, suddenly proposed, was simply the rotation of the grid system to one side. Rather than gathering all available data (observable, comparative and supplementary) for a single property before proceeding to the next, it was now suggested that observable data be recorded for all properties before moving on to the comparative process. After comparative data was acquired, the researcher could proceed to an examination of supplementary sources. It was further proposed during discussion that the conclusion categories be amalgamated to form one unit at the base of the model configuration in the interests of efficiency. Finally, the order of two artifact properties, Provenance and Function, was reversed so that Function would henceforth be considered before Provenance in each of the three question categories. This was proposed because Function entailed a greater reliance on observable data than Provenance and thus conformed to the general procedure of considering this form of evidence as early as possible within the model construct (see Table 2).

20 The new orientation appeared more logical and less rigid than its predecessor but required testing to determine if it contained inconsistencies. To that end, seminar members were once again instructed to select an artifact for analysis and to ensure, if possible, that a variety of man-made objects were represented in that selection. If the model was able to accommodate these disparate items successfully, it would demonstrate a universal and flexible application in terms of artifact research. Selections included a tea caddy, caulking mallet, lithographic prints, a pressed-glass goblet, a piece of lace, a Pembroke table, architectural drawings and a Bricklin automobile.

21 This exercise involved more research and preparation time than the first round of enquiries in the autumn and resulted in some fairly detailed papers. As these were presented, however, it became clear that the class was not in complete accord regarding the methodology. What had inspired agreement and seemed a logical structure and progression in theory did not always find adherence in practice. Some members had not followed the proposed methodology and several felt that the model remained too rigid, unduly complicating the researcher's progress by preventing freedom of movement within its structure. Others disagreed, believing that the method as developed forced the examiner to adopt an orderly approach to artifact analysis. This would permit other researchers easy access to recheck the data and conclusions and thus form a retraceable system.

22 While the concept of a retraceable analysis procedure did find agreement, debate continued over the methodology's perceived rigidity. This was manifested primarily in the fact that some had introduced documentary sources into their research before a complete examination of the artifact had been attempted. A number of seminar members felt this missed the point of the model and its indicated progression. It was their opinion that examination of the artifact by itself was essential before consultation of other sources if preconceived notions regarding objects were to be avoided and inconsistencies discovered.

23 The seminar presentation analyzing a Bricklin automobile was advanced as a case in point. It was noted that available documentation, including a Bricklin Vehicle Corporation press kit, advertised the safety aspects of the automobile.7 Yet, upon initial examination of a Bricklin, it was discovered that no handles existed on the outside of its doors to release the latches. This might present a serious problem during an emergency if the occupant were incapacitated and a potential rescuer were forced to lose valuable seconds gaining entry by other means. The seminar member who gave the presentation felt that this example appeared to justify the adoption of the analysis procedure as developed.

24 While a consensus on this issue was not established during class presentation, other aspects of the methodology were refined. A general checklist of questions in their appropriate categories continued to be developed over the course of the spring term and were applicable to a wide range of objects. The checklist was expected to be a guide only, and it was recognized that some of the questions could not elicit a response at any particular step in the analysis procedure. Thus, while the answers to some questions were commonly found during one particular level of investigation (i.e., during comparative analysis), the data to answer others might be discovered in any one of the information-gathering steps. This depended entirely on the nature of the object under examination. It also remained true, as established earlier, that the amount of information that artifacts might convey to the researcher could vary widely and that some questions presented in the checklist might remain unanswered.

25 Many of the questions encouraged the researcher to look for evidence of cultural expression in the object since this fusion had been well established early in the seminar's investigations. It was also advanced that the researcher should re-examine the artifact after the initial hands on stage, especially if new information became available through comparative data or documentary evidence that would help the analysis of the object's properties more fully. Indeed, though insistence remained that the artifact must be examined by itself during the opening steps of the analysis process, it was agreed that any single source of information could prove misleading and all available evidence should be consulted. Some felt that the methodology represented or reinforced a state of mind consistent with artifact research. Rather than succumbing to the temptation of consulting printed or written works when confronted with an unknown object, it was stressed that the material historian must develop a grammar in order to read the artifact. Use of the model would hopefully encourage this development and thus alleviate the charge of rigidity in its application.

26 As the spring term neared its conclusion, the seminar appeared satisfied that the analysis model was capable of handling a broad sample of artifacts ranging from the complexity of a Bricklin automobile to the relative simplicity of a caulking mallet. (Note: an analysis of a caulking mallet appears later in this paper.) In addition, the model formed a retraceable line of inquiry combined with an accessible overview of the entire analysis procedure. Thus, while disagreement on individual aspects of procedure remained, a general mental construct had been fashioned which, at least, could serve to reduce the researcher's preconceptions of an artifact's meaning. Its ultimate value, of course, would only be determined by continued testing.

27 The proposed analysis method encourages the historian to discard, as much as possible, preconceived notions about the artifact under study and to begin by studying the artifact itself. The investigative procedure suggests that the researcher perform a detailed examination (which may be either a written or a mental exercise) of the artifact before proceeding to other sources of information. After all observable data has been gathered, the examiner is directed to compare the artifact with objects similar to the one being analyzed. Other sources of information (documents, etc.) are introduced as supplementary data only after a complete examination has been made of the artifact and similar objects. At any stage during the process, the historian is able to re-examine the artifact, especially if new information becomes available that would help to analyze the object's properties more fully. The final step in the analysis procedure is to draw conclusions based upon all observable, comparative and supplementary data. At this point, contradictory evidence might be recognized and hypotheses formulated to explain these contradictions.

28 During each phase of the information-gathering process, the historian is seeking data from specific sources. The types of data are defined:

36 Even though, for this example, the analysis procedure has been written, the proposed analysis method may be performed as either a mental or written exercise. It is an approach for artifact analysis. The procedure is meant to encourage the researcher to examine the artifact more closely for observable data before relying on documentary sources. Naturally, the amount of information that may be extracted from an artifact depends heavily on that object (plus the examiner's observation skills and background knowledge).

37 The secret to gaining as much information as possible from an artifact rests in the interdisciplinary nature of material history studies and the observation powers of the examiner. A wide variety of relevant sources may be drawn upon to fully analyze and understand the artifact. For example, specialists from other disciplines may be able to answer questions concerning the artifact's structure or composition that few historians could answer.

38 Although an artifact of relatively simple construction (a nineteenth-century caulking mallet) is examined on the following pages, the analysis procedure has been tested on a wide selection of different objects; several were very complex. For the sake of brevity, this particular test case has been abbreviated and footnotes and a bibliography have been omitted.