Dating early 17th century French Trade Axes

Here is an excerpt of Tim and Ian Kenyon article  "The Iron Trade Axe in Ontario, ca. A.D. 1580-1650 : Exploratory Data Analysis" providing a one of the most precise dating systems for French Trade Axes from 1580 to 1650.

It is relatively well known among archaeologists, even if not well published, that the size of trade axes varies according to time and to mark number. In general, axes are thought to decrease in size through time, culminating in the light "belt" axe of the 18th century. As well, the number of marks on the side of the axe (usually l or 3, much more rarely 2 or 4) is roughly correlated to the size of an axe: the more marks, the heavier the axe.

 Despite this variation, apparent in the 77 axe sample examined here, trade axes found on ca. 1580-1650 sites in Ontario are, in essence, the same type. This becomes evident when viewed in contrast with the numerous axe varieties known in the 18th and 19th centuries (e.g. Peterson 1965). Yet as handmade products, there is significant variation : no two of the 77 axes are precisely identical in size and shape.

 Is it possible to discern subtypes or varieties within the 77 axe collection? While certain broad trends are apparent, "eyeballing" the axe drawings on the data sheets (e.g. page 9) produces only innumerable "possible" subtypes. With 77 axes and 8 variables there are 616 separate measurements to "play with". Plotting variables by time period, or against one other, yields "interesting" but inconclusive results. To hew through this welter of axe measurements, it seemed necessary to use more complex techniques for viewing the data, such as cluster analysis and multivariate statistics. These techniques, however, are complex only in a computational sense, since they may permit simpler --and more interpretable patterns --to emerge from the mass of numbers.

Figure 1: Axe Measurements

Figure 2 : Modal axe shapes and sizes for the five clusters (above) and percentage seriation by glass bead period.

Group A. The Group A axes are notable for their relatively large size, particularly in the A, B, 0, P and Q dimensions (Figure 1). All Period l axes are identified as Group A, and no Group A axes are found on later sites. Chronologically, this is the most discrete axe group.

 Group B. Group B axes are strongly associated with Period 2 sites, particularly Warminster where 9 of 10 axes were assigned to this group. Group B axes tend to be rather narrow (low IJL and N measurements). Typically these axes have only a single stamp.

 Group C. Group C axes are slightly larger than Group B axes in almost all dimensions. These axes are most closely associated with Period 3a sites (7 out of 9 specimens at both Sealey and Walker). Unlike the Group B axes, these usually have three marks.

 Group D. Group D axes are similar in size to Group C axes, but measure slightly less in overall length (dimension B) and bit width (Q). Like Group C, these axes typically have three stamps. Group D axes tend to be found on Period 3b sites.

 Group E.  Compared to the previous four groups, these are very small axes, found both in periods 3a and 3b. In contrast to Group C and D, these tend to have only a single mark. Thus they would appear to be the small size-grade of axe traded during Period 3 times.

This study has identified certain changes in axe dimensions through time, by use of cluster and principal component analysis. The Period 1 axes (Group A) are conspicuously large, representing a fairly discrete cluster. In Period 2 a smaller and rather slender axe (Group B) with one stamp becomes the dominant type. During Period 3 there are large three-stamp axes (Groups C & D) in use along with smaller one-stamp axes (Group E). The three-stamp axes change during Period 3, with length and bit measurements being reduced through time.

We should reiterate that the axe groups identified by this cluster analysis should not be considered formal archaeological types, but rather as useful, although somewhat arbitrary, segmentation of a nearly continuous range of variation.

Lastly, the use of the control sample for the assignment of specimens of unknown chronological context has the potential of identifying changing Native exploitation patterns outside of village heartlands.

Figure 3: *Recreated dating schematic with stamps, number of marks and Periods by web site author from the Kenyon's Axe Data Set by Site and Period Table.

Figure 4 : Axe Mark Type

1. Kenyon, Ian and Thomas, "The Iron Trade Axe in Ontario, ca. A.D. 1580-1650 : Exploratory Data Analysis", KEWA, October 1987