Appendix A
The Database for the Seltaman Hunting Yield

I started with a collection of 161 reported successful game capturing expeditions, recorded intermittently over a span of five years. These "hunting trips" are defined as expeditions in which one or more persons left home empty-handed and returned having killed (or more rarely simply captured live) one or more items of wild game, the news of which is then reported to the anthropologist by one or another of the hunters involved or by a person quite close to them, usually after a lapse of time. The collection as a whole has a number of sampling and quality deficiencies and thus as an entirety is of limited usefulness. But it can be converted into different subsamples that prove illuminating to the questions I wanted to answer. The questions are:

1. On average, how much meat, in unprocessed kilograms, is acquired over a fourteen-month interval through the capture of wild species? I want to integrate this figure with my fourteen-month survey of domestic pork consumption to obtain an estimate of the proportions of meat coming from all of the Seltaman primary meat sources over the year.
    
2. What proportion of this meat is furnished by the different locally available species? In other words, what is the statistical composition of the Seltaman catch? Different wild meats have different food taboos associated with them bearing upon the portion of them going to "people of women's houses" and "people of men's houses." It is only by lining up the taboos with the differential hunting yields that we can answer question C.
    
3. How much of the catch (in unprocessed kilograms) is available yearly to people of women's houses versus people of men's houses.
    

In order to answer question A, several deficiencies have to be overcome in the hunting trip reports.

1. The reports were very loosely specified as to man-hours spent.
    
2. There is almost no record of unsuccessful excursions, and, accordingly, the frequency of capture from excursions of all sizes must be extrapolated by other means.
    
3. The number of hunters in a large hunting party is often imprecisely reported.
    

Compensating for these problems to a degree, however, is the fact that from the reports, as they are itemized, a "modal hunting trip" (most common type) emerges. This is a trip taken by one to three persons for "a day" or "a night," which, when successful, yields on average 2.1 game mammals. In Seltaman bush-going practice a day usually starts in late morning and terminates no later than late afternoon, while a night usually starts after the evening meal, around 8 p.m. and terminates no later than early daylight the next morning. In table A1 the "small party" hunting, which is the modal trip, is compared to "medium" (4-10 hunters) and "large" (11+ hunters) hunting party trips.

Table A1. Hunting Party Size and Game Take

Party	Instances	Take	Average
Small	90	185	2.1
Medium	36	89	2.5
Large	29	124	4.3
Unclear	6	6
Total	161	404

Table A1 can also be examined from a different angle. One notices that the medium (4-10 hunters) and the large (11+ hunters) parties are not capturing game at a simple exponential increase over the small parties. That is because the many failed trips of small parties have not been included in the small party count (because the information was unavailable), whereas for medium and large parties the unproductive time is included. The capture rate of the medium and large parties is therefore a closer reflection of the general Seltaman capture rate. In fact, the medium (4-10 hunters) and large (11-28 hunters) expeditions can be viewed as simply clusters of small expeditions, like our modular trip. This sort of abstraction actually fits Seltaman practice quite closely. On the large-scale hunts, such as those for amyol sacrifices, the gathering of some 22 to 28 hunters first separates into large regional groups--usually three parties of 7 to 10 hunters--who will be combing a designated area for game and making camp together. When one of these subparties reaches the designated area and a camp spot is chosen, the party then breaks down into couples and triples, with an occasional single, and these are the units that then begin combing the bush for game. Thus, a 28-man hunt becomes roughly nine "small" hunting parties, and if these hunting parties stay out for four days and three nights, as is typical of amyol sacred hunts, then we can speak of potentially 63 "modular hunts" as having occurred. In approaching the matter this way, we begin to get a glimpse of the real capture rate of Seltaman hunting.

In table A2 I have broken out fifteen medium or large hunting trips for reanalysis, converting larger groups and longer time frames into clusters of modules. I have demanded, perhaps unfairly, that these hunters stay on the job both day and night, and this potentially inflates the number of modules counted. But the point is merely to show the drift of the situation.

Table A2. Larger Hunting Parties Reanalyzed

Case No.	Time Units	Pty Units	Modules	Take	Average
1	4	6	24	7
2	6	3	18	6
3	7	2	14	10
4	7	9	63	7
5	7	9	63	6
6	3	3	9	11
7	2	9	18	15
8	4	2	8	6
9	7	9	63	7
10	3	2	6	10
11	3	2	6	1
12	3	2	6	2
13	1	2	2	2
14	3	2	6	4
15	3	9	27	8
Total			306	94	0.3

The average that emerges from the reanalyzed larger hunts, that of 0.3 creatures per module, is perhaps too far on the low side. On large hunts not every small group includes an experienced hunter (though most do), not every man or boy on the trip spends time combing the bush for game, since someone has to gather firewood and build camping structures, and on long multiday hunts there may be a night or a day when the entire party is too exhausted for further hunting and simply sleeps. Nonetheless, the newly revealed average game capture per module, 0.3 creatures, is probably a closer reflection of the true Seltaman capture rate.

In general, men did not seem to hunt that frequently. Certainly, a full moon touched off the hunting imagination in the men and the frogging imagination in the women, and, were the nights clear during the three or four nights of peak moonlight, virtually every able-bodied Seltaman would participate in one or two expeditions. But at best only about half the full moons of a given year were unmarred by rain and overcast, and there were sometimes social reasons (e.g., a recent death in the village) to forgo hunting even under ideal conditions. The deficit had to be made up by the more private solitary or small group (man and boy, two men, man and wife) excursions that occurred sporadically during other times of the month. Judging from my four closest informants--all of them accomplished hunters--these occurred one to three times per month. All things considered, we can speculate that if, over the course of my fourteen-month record, the fourteen reasonably able hunters in the village for which I kept this imperfect meat record, each made two modular trips per month ([2 x 14] x 14 modules) and participated in four large-group hunts (63 x 4 modules), the fourteen-month catch would be approximately 194 creatures (644 modules / 0.3 creatures).

Now for the question of the statistical composition of the Seltaman catch of midsized game mammals. The total sample of 161 hunting trips was dominated by accounts of hunting expeditions with Bobsep, a fabulous hunt dog that had just been put down prior to my visit in 1993. Yet data from my first and longest visit indicated that at most only a third of all hunting took place with a dog as a hunting companion. Furthermore, most dogs were not as good as Bobsep, and most dogs did not specialize to Bobsep's degree in the capture of kwemnok (P. gymnotis). It thus became necessary to create a subsample from the total that was more representative in terms of dog-versus-nondog hunting and more representative of the contribution of the average dog. This is the first adjustment carried out. It was pointless as well to include in this representative subsample any trips for which the game catch was unspecified as to type, since we are concerned here with the composition of the catch. And yet, because so many of the trips without Bobsep, having come from my earlier research, were inferior regarding the specification of what was caught, I found myself left with a numerically shrunken subsample once all poorly specified trips (i.e., no specification or only partial specification) were eliminated. As a compromise, I have included the partially specified trips along with the totally specified ones. But the percentages of the different game creatures are based on the total of specified game only. What we have, then, as a representative sample of Seltaman hunting is a collection of 91 trips in which 30 percent of the trips (27) involved a dog, and one-tenth of the total number of trips (9) involved Bobsep. Bobsep's 9 trips were randomly chosen. Table A3 shows the game composition of the representative subsample (sample 1); the game composition of the subsample of all specified game (sample 2); and the game composition of the subsample of all trips in which the game from the entire trip was specified (sample 3). The latter two tables include trips with Bobsep and trips with other dogs.

Table A3. Hunting Yield in Three Subsamples with Capture Frequency Rankings

			Sample 1 Representative (181)		Sample 2 Game specified (326)		Sample 3 Trip specified (303)
Rank	Type	Species	Number	Percentage	Number	Percentage	Number	Percentage
1, 2	watom	D. vanheurni	41	22.65	73	22.39	69	22.77
2, 1	kwemnok	P. gymnotis	27	14.92	103	31.60	97	32.01
3	ngarem	P. vestitus	26	14.36	40	12.27	37	12.21
4	kitem	P. sericeus	22	12.15	22	6.75	20	6.60
5	kayang	Ps. cupreus	14	7.73	19	5.83	16	5.28
6, 7	sop	Ps. forbesi	11	6.08	11	3.37	7	2.31
7, 6	deim-arik	P. orientalis	10	5.52	16	4.91	16	5.28
8	kuter	Uromys sp.	7	3.87	7	2.15	7	2.31
9, 10	ngarfem	Dact. sp.	5	2.76	5	1.53	5	1.65
10, 11	kayuuk	M. longicauda	5	2.76	5	1.53	5	1.65
11, 13	akhuni	D. goodfellowi or D. spadix	3	1.66	3	0.92	3	0.99
12	kimisok	P. raffrayana	3	1.66	4	1.23	4	1.32
13, 15	ngorim	P. carmelitae/ P. matinam	2	1.10	2	0.61	2	0.66
14, 8	bakonkaak	E. kalubu	1	0.55	7	2.15	7	2.31
14, 16, 19	mein	Mallomys sp.	1	0.55	1	0.31	0	0.00
14, 16	sarip	P. maculatus/ rufoniger	1	0.55	1	0.31	1	0.33
14, 13	somin	H. goliath	1	0.55	2	0.61	2	0.66
14, 15	wares	A. imitator	1	0.55	1	0.31	1	0.33
15	yakheil	Z. bruijni	0	0.00	1	0.31	1	0.33
16, 12, 13	aboysep	D. albopunctatus	0	0.00	3	0.92	3	0.99
	Total		181	100.00	326	100.00	303	100.00
	Unspecified		66		0		0

After assessing the representative subsample for the statistical representation of captured species, I returned to the larger sample to compare the species capture rates of trips with a dog present against all other trips. The results are represented in table A4.

Table A4. Subsample of Captures from Trips with Dog Present Versus Trips with Dog Absent

		Dog Present		Dog Absent
Game Type	Species	Number	Percentage	Number	Percentage
watom	D. vanheurni	24	34.30	17	15.30
kwemnok	P. gymnotis	21	30.00	6	5.40
ngarem	P. vestitus	5	7.10	21	18.90
kitem	P. sericeus	3	4.30	19	17.10
kayang	Ps. cupreus	4	5.70	10	9.00
sop	Ps. forbesi	2	2.90	9	8.10
deim-arik	P. orientalis	4	5.70	6	5.40
kuter	Uromys sp.	0	0.00	7	6.30
ngarfem	Dactylop sp.	0	0.00	5	4.50
kayuuk	M. longicauda	0	0.00	5	4.50
akhuni	D. goodfellowi	3	4.30	0	0.00
kimisok	P. raffrayana	0	0.00	3	2.70
ngorim	P. carmelitae	0	0.00	2	1.80
bakonkak	E. kalubu	1	1.40	0	0.00
mein	Mallomys sp.	1	1.40	0	0.00
sarip	S. maculatus	1	1.40	0	0.00
somin	H. goliath	1	1.40	0	0.00
wares	A. imitator	0	0.00	1	0.90
Total		70	100.00	111	100.00