publisher colophon

4How to Trap a Fly

At the end of the pan-African Agricultural and Veterinary Conference in Pretoria in August 1929, Chief Entomologist Rupert Jack visited the KwaZulu area affected by mpukane. He spent ten days with R. H. Harris, the man in charge of tsetse operations there, learning about the Harris flytrap then in trials. In November of the next year, Jack again returned to KwaZulu to witness its demonstration. As he put it, the traps “function[ed] remarkably” well and the demonstration was “decidedly impressive.”1

The Harris flytrap experiments began in the Nyambani valley of Kadoma in 1930, with various modifications alongside other models to determine what was best for Southern Rhodesia. Each was painted in a different color to determine which one yielded the most mhesvi catches. Dark blue and black turned out to be the most attractive, while results on other colors proved “disappointing in many respects.” (Potts 1930; Nash 2007; Lloyd 1935; Swynnerton 1933; Jack 1939; Barrass 1960). Effectiveness varied considerably in the twenty-two traps according to position, amount of movement within the vicinity, seasonal and daily weather conditions, and mhesvi concentration. Most of the flies were caught from August to September, with the numbers decreasing in October and becoming “almost negligible” as the rains started. By contrast, a single mufrayi caught twice as many flies with a net in one hour as the best trap caught in a month.2

In 1932, Jack emphasized that any significant catch only took place during a small portion of the year “and even then has been too small to afford promise of practical benefit from the use of the traps.” As he put it, “The reason for the failure of these fly traps in Southern Rhodesia lies deeper than in details of design—it lies, in fact, with the principle itself.”3 Chorley explained in 1937: “Any trap depending on shade as an attractant must give poor results with G. morsitans which is an open savannah loving fly, not so dependent on shade as G. palpalis or G. pallidipes.”4 Precisely the difference with the KwaZulu experiment! The trap could only be effective with a thorough knowledge of the bionomics of mhesvi. Deployed against mhesvirupani in KwaZulu, the Harris trap was very effective, but it was generally ineffective against mhesvirutondo elsewhere in Africa.5

K. R. S. and M. G. Morris’s investigations in the Gold Coast (Ghana) in 1949 yielded a trap that would become significant in Southern Rhodesia. Fewer examples illustrate the enmeshment of vachena’s means and ways of controlling mhesvi, on one hand, and vatema’s ways of life and spiritualities on the other. In 1949, the two researchers sought “a means of controlling G. palpalis and G. tachinoides in sacred groves,” pernicious foci for the concentration and dissemination of mhesvi and n’gana and frequently visited by local vanhu. The investigators defined the problem thus: “Rigorous taboo opposes the cutting of any vegetation within them, and one or two early attempts to introduce clearing methods resulted in the loss of confidence and cooperation of the people” (Morris and Morris 1949, 494; Morris 1950). The methods—the means and ways—had to be tailored around the cultural dynamics. “Thus,” Morris and Morris (1949) concluded, “J. K. Chorley’s Crinoline trap might be mistaken by a tsetse for the body of a man; small forms of Harris or Swynnerton’s Screen traps [developed in Tanganyika] might represent the local dwarf breeds of cattle, but both were too big to resemble what are by far the most numerous of the food hosts along the banks of rivers: sheep, goats and, in the more thinly populated parts, duikers” (494).

Morris and Morris therefore designed a new “animal trap” (see figure 4.1a, b) resembling a mbudzi (goat-)-sized animal, which they trialed alongside Chorley’s eighteen-inch diameter, twenty-four-inch long Crinoline traps made out of khaki and black material; Harris traps two-thirds standard size; and Swynnerton’s awning screen box (ASB) traps with screens 4 ft. wide, 2 ft. long, and 2 ft. deep (Morris and Morris 1949, 494). The attractant was “a cylindrical body, 2 feet long by 1 foot in diameter, standing with the top or shoulder 2 feet from the ground, and covered with ordinary light brown hessian or burlap” (Morris 1961, 905).


Figure 4.1a, b The Morris folding trap, showing its removable parts for ease of carriage (top) and the fully assembled trap (bottom).

Sources: Morris 1961, 907 (top image) and Morris and Morris 1949, 495 (bottom image).

By 1960, many traps had been designed (Swynnerton 1933, 1936; Chorley 1933; Lewillon 1945; Morris and Morris 1949; Morris 1960). All showed that a trap that worked well against river-hugging mhesvirupani (vachena’s G. palpalis and G. pallidipes) might perform badly against savannah-loving mhesvirutondo (what they called G. morsitans). Both Swynnerton’s ASB trap and Langridge’s box screen trap rated higher than the Morris trap (Moloo 1973, 231)—quite the opposite of mhesvirutondo-dominated Rhodesia!

The interest in traps seemed to have died down in the immediate postwar period and the 1950s. However, as game elimination and concerns about OCP toxicity gathered pace, BTTC revisited the possibility of trapping mhesvi, especially in those districts where its threat was significantly reduced and where traps could act as surveillance apparatuses. Two mechanical traps were considered, based entirely on the artificial refuge model. One was the Morris trap, with a body of black cloth and a black screen just like Swynnerton’s trap. The other was a Manitoba fly trap with a decoy composed of a black ten-gallon drum mounted fourteen inches above the ground (Thorsteinson, Bracken, and Hanec 1964).6

The idea of trapping mhesvi had seven selling points, enunciated in the BTTC’s annual report of 1982. First, it integrated control and survey operations otherwise performed separately, thus saving costs. Second, it was a simple means for financial-resource-starved Africa. Third, it reduced the heavy demand placed on foreign currency, often associated with OCP spraying. Fourth, it placed only steady, constant demands on labor and equipment all year round, unlike OCP spraying’s peaks. Fifth, it guaranteed a reasonably constant effect on mhesvi for the whole year, compared to the sharp but short pressure from spraying operations. Sixth, it allowed sufficient time to correct any errors without jeopardizing the entire program, unlike spraying cycles. Finally, compared to OCPs and “game elimination,” it was an ecologically clean and friendly method.7

What follows is a discussion of the types of baits that were under serious scientific, technical, and economic consideration, each categorized according to the attractant or stimuli used and how it handled the mhesvi, the method of treating the attracted flies, and the costs and efficacy per each unit deployed. The underlying principle of a trapping system was that mhesvi had very small reproductive potential, such that a slight reduction in the chipukanana’s reproductive rate or increase in its mortality rate was enough to control its entire population. This could be done by attracting the mhesvi to artificial baits laced with killing or sterilizing agents.

Attracting the Fly

The role of mhuka as attractants had received attention prior to its escalation in late 1940s Southern Rhodesia; it was the key to why some traps succeeded for one type of mhesvi but not another and why a one-size-fits-all approach was deemed a terrible idea (Swynnerton 1921a, 1936; Harris 1938; Jack 1941; Vanderplank 1944).

From 1957 on, the most significant attractant research was performed at Kariangwe and then, from 1965 on, at Rekomichi Field Research Station. The research sought to establish the best attractants of mhesvi. The baseline for the investigation was quite clear: Mhesvirupani was not easily attracted to vanhu but was attracted to mombe, leading to the need to “compare catches made by various methods at the same time over the same territory.”8 The research focused on sight (visibility) and smell (olfactory) as attractants.

Visual Attractant Research at Rekomichi and Sengwa

The most interesting studies on visual attractants were conducted at Rekomichi (Rekomitjie) from 1966 on, with subsidiary research near the Ruyese River and at Sengwa. The experiments at Rekomichi began in August 1966 and ended in December 1967. After that, the research continued at Sengwa until October 1970, while the Ruyese study was conducted for only two weeks in September–October 1967 (Vale 1971, 331).

You may remember that vatema had long observed that mhesvi tended to be attracted to black forms, especially those resembling mhuka.9 Old Mabuzana said as much to Swynnerton. The reason for moving at night was, in part, to neutralize blackness as a factor in visibility and render everything the same to mhesvi. The Rhodesians were working with the opposite dynamic; instead of trying to prevent mhesvi from biting, they deliberately deployed black forms during the day to flush them out.

For the Rekomichi experiment, five artificial refuges were found to attract mhesvi, all of them with a dominant black component. The hut refuge was composed of a hut measuring 6 × 6 × 6 in., with walls of sacking and clay mud covered by a thatched roof, all designed for mhesvi to come in and rest. The trellis refuge was a 10 × 8 in. wooden trellis, with a lean-to grass-thatch roof over a drum covered with soil for the same purpose. The box refuge was a forty-four-gallon drum with one end cut out, placed horizontally on the ground, and covered with soil to insulate against the sun’s rays.10 The wigwam refuge was a thatched, triangular structure offering shade over the entrance to a black insulated ten-gallon drum. The last refuge was a set of pipe refuges made of concrete pipe, 4 ft. long by 2 ft. in diameter, placed horizontally or vertically, and insulated with a 1.5 in. layer of thatch (Vale 1971, 333; see figure 4.2 and figure 4.3 for a closer view).


Figure 4.2 Five refuge models at Rekomichi and Sengwa in 1971.

Source: Vale 1971, 332.


Figure 4.3 A closer view of the flask, wigwam, and box traps.

Source: Vale 1971, 334.

Each refuge was created within and applicable to three different vegetation types. Two variations were made—namely, the location and size of the drum and the materials used as cover (thatch, sacking, or black cloth). These changes had little effect on the number of flies attracted. More useful was the addition of a curtain that could be closed at a distance to keep out any flies that might have followed the observer, and a drum refuge site developed as a trap that allowed mhesvi to get in and rest, but kept them hostage.11

The “glossinologists”—as vachena specializing in studying Glossina began to call themselves—studied the composition of refuge site samples, focusing on the physiology of the flies caught. They compared these with samples from Morris traps, a tethered bait ox, and man-only bait.12 Fifty percent of mhesvi caught in refuge sites had “incompletely digested blood visible through the abdominal wall.”13 The diurnal pattern of refuge behavior was of especial interest. Attractants like tethered oxen and traps were inspected at about 1400 hours, when the sun was blazing hot.14 Seasonal patterns of refuge behavior were also studied, with catches peaking in late September and early October (the hottest period), gradually declining as the rains began, hitting their lowest in late December and early January, and staying generally low through to May.15

To assess the influence of refuge site positioning on the quantity of catches, over fifty drum sites were plotted within a one-square-mile area containing varieties of vegetation. Under observation from October to November 1966, one or two mhesvi were caught per day in the refuge sites in the open mupembere or mubondoroko (vachena’s Combretum) bushland, on musasa, munhondo, muunze, and mupfuti (all vachena’s Brachystegia) and other woodland on the Zambezi escarpment slopes, and mupani or musharu (mopane in setswana) woodland at its base. Moderate catches of up to fifty were obtained daily from areas with some shade and adjacent to steep banks, particularly gullies fringed with leafless thicket. Very good catches, frequently two hundred mhesvi, sometimes over four hundred, were obtained in deeply shaded areas, especially among leafy, riverine musika (Tamarindus indica) trees. How then could they explain “the great differences between the magnitude of catches from exposed situations and sheltered situations?”16

By the end of 1967, BTTC had successfully developed the refuge trap to catch the chipukanana as it sought sanctuary from prey or the elements or sought to rest. It was thus designed for resting sites and stationary objects. The centerpiece of the Rhodesian approach to refuge traps was the Morris trap described earlier.

The success of the refuge trap concept led researchers to believe that the use of mobile models of mhuka could help flush out mhesvi and expose the role the distribution and behavior of mhuka played in mhesvi’s host preference.17 What the BTTC called attraction studies were composed of observations on numbers and behavior of mhesvi within the proximity of live mhuka and the development of mobile and stationery models as devices to explain these observations. This experiment was undertaken at Sengwa Research Station. The timider mhuka like njiri and nguruve, and zvipfuyo like imbwa and mombe were placed into one group, while the fickler ones like mhara and dzoma were placed in another. Mhesvi freely landed on the former, much less so on the latter.18


Many mhesvi landed near the eyes of the adult njiri, fewer on the adult nguruve’s, and none on those of the other mhuka. Large numbers engorged on njiri (warthogs), especially within two inches of its eyes. Mhesvi probed puppet of njiri, nguruve and dzoma stood next to live adult njiri, but not near the eyes. The conclusion: It was possible that olfactory stimuli played a role in mhesvi’s responses to njiri’s eyes.19

By 1972, it was now known throughout the tsetse research fraternity that mhesvirutondo usually fed upon hosts resting in fairly dense vegetation, where smell rather than sight seemed a likely attractant. Yet to be determined was the role of olfactory (odor) stimuli in host finding. Male mhesvirutondo proved more responsive to moving targets than females; the females preferred targets whose movement was interrupted (Gatehouse 1972). There was strong belief that olfactory attractants could improve mhesvi population survey techniques and give a new dimension to the means and ways of controlling mhesvi if deployed with pesticides or chemical sterilants (Turner 1972, 25).

In the 1972 field trials, the ox attracted the most flies, but large catches still were made with mbudzi (goat), mubhemhe (donkey), njiri, hwayi (sheep), nguruve yemusango, nhoro (kudu), mhara, and nyati.20 Two hundred and twenty-five flies were caught in portions of cage and netting without bait, 442 in those with visual bait alone, 2,730 with just odor bait, and 3,868 with a combination of visual and olfactory bait. That translated to a ratio of 1:2:12:17. This compared with sixty-six for the small electric net without bait, 803 with just bait, 892 with odor bait only, and 2,902 with visual plus olfactory bait. The ratio was 1:12:14:44. These results led to two important observations regarding net size: First, odor alone was as good as visual stimuli alone when small nets were used, but “odour was six times more effective than visual stimuli when the large net was used.” Second, adding visual stimuli to odor “increased catches only by half when the large net was used, but trebled catches when the small net was employed.”21

Some of the nets were deployed to catch mhesvi on their own without any ox or person around. In its mounted position, the blue device would look like an ox; mhesvi would investigate this “animal,” fall in, and get caught.22 Without bait, catches were evenly distributed, but with the model as bait, catches were concentrated on the cage. With odor-only bait, the concentration on the cage was less marked than with the model, showing less precise orientation. Instead, catches were strongly concentrated on the downward faces, indicating an upwind flight to the baits. Finally, with both model and odor as bait, concentration was pronounced, both on the cage and on the downwind faces.23 Two observations were made: First, few flies ranging meters off the bait could be attracted to a stationary visual stimulus; second, many flies were attracted by olfactory stimuli from afar, but needed visual stimuli at close range for precise orientation. The verdict: Odor alone had poor efficacy, even when used with small nets.

No significant difference was noticed between odors from a normal ox, one with a dampened coat to boost body odor, and yet another with its body thoroughly washed in water or a Teepol solution to decrease odor. To determine the importance or insignificance of body odor, a special pit was dug and split into two similar chambers using an airtight canvas sheet. The mombe was isolated in one chamber, with its head and upper neck passed through a laced slit in the sheet and isolated in the other chamber. The tests showed head odor yielded catches “several times greater” than body odor and indicated that head odor was responsible for mhesvi’s attraction to mombe.24 Explaining the concept with a picture portraying an African field assistant pointing out the ventilator, Vale said: “To study attraction to stationary sources of host odour, an animal is hidden in a roofed pit and air from the pit is blown out at ground level through the opening to which the man is pointing. Flies attracted to the odour are caught by the electrocuting net suspended within the rectangular frame” (Vale and Phelps 1974, 1; see also figure 4.4). Air from pits was blown out at ground level close to an electric net to trap all attracted flies. Although the odor of ox was quite effective, large catches were still obtained using mubhemhe, mbudzi, hwayi, dzoma, nguruve, and nyati. Munhuwi wemunhu (human odor) proved ineffective and reduced the effectivness of ox’s odor when deployed in combination (Vale 1973).25


Figure 4.4 Roofed pit with a jon’osi inside to generate odor.

Source: Annual Report 1974. SACEMA/TA.

In 1975, Vale determined that flies were attracted to a stationary visual target that released odors; the greater the odor, the greater the number of catches. The experiment deployed a ventilated pit like the 1974 one, but sited in semievergreen woodland on the Rekomichi floodplain. It measured 6 m long × 4 m wide × 2.1 m deep, with three 15 cm ventilation shafts equipped with an electric extraction fan 12.5 cm in diameter to ensure adequate evacuation of odor from up to six oxen. Finally, a black visual target (a cylinder 50 cm long by 37 cm in diameter) was placed 37 cm above the odor outlet. A 90 × 90 cm electric netting was placed downwind of the target to catch all attracted mhesvi (Vale and Hargrove 1975, 46; Vale and Phelps 1974).

The researchers were expecting at least two things: First, they considered excluding visual targets because they were interested in odor targets, but used them anyway because they increased catches “by concentrating near the net those flies initially attracted by odour.” Second, they expected that the electric net would not cope effectively with high numbers of flies attracted by six oxen (Vale and Hargrove 1975, 47–48).

There is no record of experiments for 1976–1981, because Rekomichi fell within a hotly contested warzone between varwiri verusununguko/abalweli benkululeko (freedom fighters) and the Rhodesian Security Forces, but we catch up with BTTC researchers after independence, while they were isolating the effective ingredients of the odor of mombe to dispense it artificially. Carbon dioxide and acetone had now been identified as the chemicals at the center of the smell. Applied in doses equal to the odor of mombe, they yielded half the number of flies the live bovine produced, suggesting more components in the odor of mombe beyond just carbon dioxide and acetone. The race was on to isolate these components and explore their efficacy in lieu of carbon dioxide, which was “expensive and inconvenient to dispense in large-scale operations.” That is how researchers found that “acetone alone, at economical doses of 5–500 mg/ha, could increase trap catches by 2–5 times.”26

Several chemicals were thus identified and tested in both built lab and field, but just one—1-octen-3-ol—was subjected to intensive initial tests. When added to carbon dioxide and acetone, it accounted for virtually all the efficacy of natural ox odor (hwema) against mhesvirutondo, but just two-thirds of that against mhesvirupani. The chemical 1-octen-3-ol was effective alone or with acetone; the problem was that it became a repellent if dispensed in a high dose, unless diluted with air prior to exposure to flies. An air-mixing fan was used; alternatively, the odor was simply released one meter upwind of the trap so that the chemical would be diluted by the time it reached the trap. With these facts now mastered, the next step for BTTC was to establish tsetse behavior in odor plumes.27

By 1984, the odor-identification project had isolated four attractive elements found in host odors—carbon dioxide, acetone, butanone, and 1-octen-3-ol—each capable of boosting mhesvi catches several times over. Further attractants in ox odor remained unidentified; completing their identifcation was “likely to take a long time.” The dung and urine of nyati seemed nonattractive, but catches from traps increased two to four times “by placing next to them sacks used as bedding by a bushpig.” The attractive component was found to persist on the sacks for five to six weeks in a nonacidic form readily soluble in chloroform.28

The samples of the chloroform washings were sent to the Tropical Products Institute (TPI) in London and the Tsetse Research Laboratory (TRL) in Langford, Bristol. The process involved collecting host odor extract in Zimbabwe; fractionating and chemically identifying these extracts into their constituents at the TPI; lab studies at TRL to determine whether the chemicals the TPI had identified elicited a suitable behavioral response from mhesvi; and field studies in Zimbabwe to determine whether the fractions or known chemicals identified at the UK labs to produce suitable responses under built laboratory conditions could produce the same responses in the field. Thus, the trap activity on the ground was also geared toward the branch “meeting its obligations” to the collaborative project by refining its field methods for studying the response to odors.

In 1985, it emerged that there were at least three other as-yet unidentified groups of components—namely, an ox odor that was very volatile when passed through a charcoal filter; a rather involatile one from nguruve present on sacks brought into contact with these wild swine; and another involatile one, mombe urine. The ox odor was the distant attractant of mhesvi to hosts and expressed “marked increase” of catches with every dosage increase. Nguruve and mombe urine fractions were principally short-range attractants, especially responsible for boosting zvipukanana’s readiness to enter traps once visually attracted or odor-attracted. The fractions achieved a fivefold increase as attractants, but had a saturation point beyond which no further increases in catches could be obtained. In fact, sacks and urine were found to stimulate chemoreceptors of mhesvi, some of them repellents (small fatty acids at high dose) or neutral (cresols at various doses). Some were attractants, like a p-cresol sample the BTTC sent to Burkina Faso for comparisons.29

The discovery that carbon dioxide was necessary to interest mhesvi in the trap in large numbers significantly reduced the value of using catch-and-retain and sterilizing traps versus targets with OCP-treated netting. It was already known that only twenty cubic centimeters per minute (cc/min.) of carbon dioxide in place of 2 liters per minute (l/min.) present in ox odor was enough to entice mhesvi into traps, yet even such a low dose was too expensive. In lieu of carbon dioxide, BTTC resolved to modify the trap design or find a substitute for carbon dioxide while using targets with netting to control mhesvi in the meantime.30

All this research should bring to mind an earlier moment in history, when vanhu vatema burned dung indoors and underneath their zvipfuyo and smeared it on the skins of mhuka to ensure their safety against mhesvi.

Sticky Traps

I grew up trapping and catching shiri—lots of them—with birdlime. We made the birdlime ourselves. In Zimbabwe, boys tap rubber from at least three trees: chitatarimbo (which vachena now call Euphorbia matabelensis; see figure 4.5), mukonde (Euphorbia ingens), and mutsamvi (Diplorhynchus condylocarpon). In our village, we had plenty of zvitatarimbo (plural of chitatarimbo) growing wild in the forest or as a hedge in people’s wetland gardens. Kugumha hurimbo (harvesting birdlime) is an old tradition, practiced for generations by our ancestors and continuing among youths today.


Figure 4.5 Chitatarimbo/Euphorbia matabelensis.

Source: Photo by Lewis Mavhunga, Mavhunga Village, May 2017.

We would cut across the chitatarimbo’s trunk using knives to extract the white milk used to make hurimbo, the sticky rubber for trapping shiri. Two apparatuses were required: a knife and a small container. We would make a small, sharp knife by simply beating a three- to six-inch nail with a stone or claw hammer then sharpening (kurodza) the blade on a rough granite rock. The container was usually the base of a shoe polish tin or a disused canned jam or baked beans tin cut in half. We also carried a small amount of Vaseline Blue Seal petroleum jelly or Olivine cooking oil—an ingredient to make the rubber less stiff, more tactile. Each of us would collect his own hurimbo from the chitatarimbo. When enough was gathered, we added a little oil and put our small tins on a small moto. When the rubber showed signs of being well cooked, we took the tin off moto and stirred the mixture, either with the knife blade or a small stick. The substance now wound into a ball in the center of the tin, around the tip of the blade or stick. We then chewed it, spitting the saliva containing the excess juice, and stretched the birdlime in short pulls, bringing the two ends back together again, then stretching, until the hurimbo was well mixed.

Then, we were ready to trap. We wound the rubber around a specially selected, thin, and nontoxic stick called mudziti, which we placed atop the tree canopy where shiri rendezvoused en route to distant places or when helping themselves to mhunga (millet) fields, or we placed midziti (plural of mudziti) by the zambuko (drift), where shiri came to drink and bathe. Once a shiri landed, its claws wrapped firmly around the stick, it was stuck—but as it flailed its wings to make good its escape, the feathers also became stuck. As soon as it landed, we were already running to collect it. Two dzimbahwe proverbs capture the stickiness of hurimbo that we experienced as boys: First, “Poverty is birdlime, it sticks.”31 Second, “The bird stuck to the birdlime is the one that says tsviri-tsviri; that in the mouse trap says the sky has fallen” (Hamutyinei 1992, 40).32 Hurimbo caught shiri alive.

One of the biggest surprises I encountered during research for this book was vachena’s appropriation of hurimbo to trap mhesvi. The ivory poacher Cecil Bvekenya Barnard, whose story is told in Transient Workspaces, built upon the local art of making traps laced with adhesive rubber from zvitatarimbo and other rubber trees to catch shiri in order to make his own brand of anti-ndedzi trap. He says he never lost a single mubhemhe to ndedzi in fifteen years of poaching in Hlengwe country between 1914 and 1929. His concoction was composed of animal fat, Cooper’s dip, tree gum, and honey, which he smeared all over his donkeys. Any mhesvi that landed became stuck to the gum, and the arsenic poison would finish it off. “The donkeys certainly became pretty sticky-walking fly traps,” but they survived and confirmed the concoction’s efficacy (Bulpin 1954, 58).

The idea of turning zvipfuyo into mobile sticky traps did not catch on in government circles until the immediate post–World War II period. At that point, Southern Rhodesia’s veterinarians and entomologists started applying DDT to cattle-dipping liquid chemicals to kill mhesvi as it landed and attempted to feed. The major concerns were that the animal sprayed might succumb to skin lacerations or hair loss or lick its skin and poison itself. The latter problem was resolved by concentrating the application around the neck area, the most olfactory part of the animal’s body unreachable by its mouth.

Two experiments brought together OCPs and hurimbo as ingredients for the transformation of an ox into an automobile trap. One produced satisfactory kills of mhesvirupani when treating mombe with a solution of 9 percent DDT in oil twice a week—a very high concentration. A second mixed DDT in gum and ox serum, which was applied against another mhesvirupani. Three other studies involved mixing DDT in an adhesive made of coumarone resin. Yet another study used boiled linseed oil as an adhesive solvent, which, when dry, did not produce DDT crystals—thus resulting in low toxicity and therefore ineffective (Wilkinson 1948).

These researchers persisted with experiments because DDT-treated oxen had been successful against horn flies in the United States. Could mhesvi be exterminated by flooding their habitat with DDT-treated oxen that outnumbered mhuka? Could mhesvi be enticed to prefer mombe? Some researchers argued that because mombe were quite favored by mhesvi, they could, when injected with the trypanicidal drug phenidium compound, be kept in the mhesvi-infested bush longer, with the DDT preventing further bites (Vanderplank 1944). In Texas, Florida, and Kansas, ranch cattle dipped in or sprayed with 1 percent DDT aerosols, emulsions, and suspensions had kept cattle free of horn flies for weeks. Why, it could also work here!

Whiteside’s experiment at Old Shinyanga in Tanganyika summarizes the method for treating oxen with DDT: “A solution containing 9 per cent. w/v pure DDT and 9 per cent. w/v resin in commercial groundnut oil was used. … The groundnut oil was heated to 105–110º C while dissolving the DDT and resin, and the resultant solution applied (cold) to oxen at the average rate of 110 cc. per ox, excluding wastage, corresponding to 9.5–10 grams pure DDT per ox, or about 450 mg. per square foot of body surface. The head alone was left unsprayed” (Whiteside 1949, 124). Yet, as Whiteside would discover in 1949, mhesvi in Africa was very different from horn fly in the United States: “Extermination—not merely ‘control’—is required; the contact of tsetse with cattle is relatively fleeting—perhaps only 1 minute in four or five days—compared with that of horn fly; and whereas against the latter it is sufficient to treat the heads and backs of cattle, against tsetse the legs and belly are most important, and are difficult to spray efficiently” (123).

Could an efficient method of spraying be found? Not at the time. By 1949, none of these or other follow-up experiments had succeeded (Symes et al. 1948; Whiteside 1949; Woodcock 1949; Burnett 1954; Burnett, Robinson, and Leroux 1957). Simply because it had succeeded in the United States or Britain did not mean it would succeed anywhere else.

This was also what the Division of Entomology had found in Southern Rhodesia in 1947. The entomologists were adding DDT to dipping fluid so that as mombe plunged into dipping tanks and drenched their bodies, the emulsion would soak in and leave a residual effect on the skin as it dried. When a mhesvi landed upon it to feed, the OCP would kill the chipukanana through skin absorption. This chemical was applied to the dip mix at Muumbe dip in the southeast in 1947 to give an oral prophylactic to mombe in the face of a mhesvi onslaught from Portuguese territory. The view of the vet, entomologist, and Native Commissioner was that DDT could insulate zvipfuyo from mhesvi’s bite. However, the mix was too strong: “Certain of the ingredients of the emulsion caused an apparent scalding and a violent reaction resulting in temporary paralysis of the hind-quarters, and after two subsequent attempts, the project was abandoned.”33

In any case, the rationale for using chepfu (poison) was to effect the mass destruction of mhesvi, rather than using single oxen moving around as bait, which would soon become clogged with trapped flies. That method was wasteful and extremely toxic to oxen. The idea was not abandonded, however. In 1965, BTTC and DNPWLM looked for a commercial insecticide that could “safely be applied to the coats of mombe in quantities which are toxic to tsetse flies”34 and kill mhesvi within seventy-two hours, if the chipukanana was exposed for two minutes. The second objective was to find out how far mombe could tolerate the pesticide on their skins, so as to determine what concentrations “could be tolerated by them and yet lethal to tsetse.”35 The critical time to poison mhesvi was when it settled on the body of the animal to draw its blood meal, but the poison stayed on the body of the animal much longer.36 The problem was that some of the OCPs, when applied to coats of mombe, “caused a depression in the production of cholinesterase that threatened to be lethal with further applications.”37 The holy grail would be to “safely incorporate an effective insecticide into the cattle dips used in routine tick control.”38

Mhesvi became a bonanza for Big Pharma. By 1968, well over fifty insecticides had been tried as “screens” on the hides of mombe.39 One of the manufacturing companies involved was Cooper Technical Bureau (CTB), a UK company based out of Berkhamsted and famous for making Cooper’s Dip, an antitick dipping chemical for hwayi and mombe. CTB submitted to the Tsetse Department twelve insecticides for potential use in that role, including asuntol, neguvon, imidan, and 42/1/65. However, when these specific four were tested, they proved to be insoluble in water. Another company profiting was the petrochemical giant Shell Chemicals (UK), which supplied birlane (applied as a 24 percent solution). CTB changed its name and operated as Cooper, McDougall and Robertson (CM&R) until 1973, when it became Wellcome. As CM&R, it supplied sevkol (30 percent), ethion (30 percent), simuthion (60 percent), and DBM (38.63 percent). The compounds alodan, Dow E.T. 57, ruelene, bromophos, and bromodan were also mixed in a solution of eighty parts of petroleum ether and twenty parts of olive oil.40 In 1969, twenty-two compounds were put on trial.41

We can now rejoin hurimbo. That same year, the branch announced that a 1:1 mixture of polybutene, manufactured by Klipfontein Organic Products Corporation of South Africa, and a hurimbo (gum) obtained from chitatarimbo could form an “effective sticky deposit” for trapping mhesvi. What was left now, it said, was “to find an effective sticky substance which could be mass produced from commercially available raw materials.”42

It is significant that sticky techniques were now being integrated with attractant-research-based Morris traps to produce a new kind of trap. A synthetic rubber mixture of 9:1 polybutene to Adlatex (made locally by Glue and Chemical Products, Salisbury) had proved “simple to prepare” and maintained “a high level of effectiveness almost indefinitely.”43 Vale, the glossinologist in charge, felt that the color of the sticky substance was critical to “the efficacy of attractants.” It had to match the color of materials upon which it was applied and that mhesvi favored (Vale 1969, 1).

Five attractants were deployed. The first was a Morris trap with a “body” of black cloth. The second had a black body plus a sticky screen on the inside of each “flank.” The third had a black body and sticky screen hanging vertically inside the body of the trap. The fourth also had a black body and sticky screen arranged just like the third, but with a sticky screen on the outside of each flank. The fifth was a ten-gallon drum covered with a black cloth and mounted horizontally fourteen inches above the ground, resembling a cow. On each flank was a sticky screen. All sticky deposits were colored black. “It appears,” Assistant Director Gerald Cockbill remarked in his 1969 report, “that many more tsetse visited Morris traps than were trapped by the cage recovery system. Since the sticky screens inside caught so few tsetse, it seemed that the cage recovery was inefficient because tsetse did not enter the body of the trap.”44

Further experiments were undertaken to compare the recoveries from three scenarios: a sticky model alone; a sticky model plus munhu mutema (black man) walking three feet away; and a nonsticky model accompanied by munhu mutema with net in hand, catching flies landing on the model. The models were all pulled by hand. The unaccompanied sticky model yielded more male and female mhesvi than the nonsticky model. The entomologists drew one conclusion: “Since the model plus man combined with sticky recovery gave catches intermediate between those obtained from the sticky model alone and from the non-sticky model plus man incorporating hand-net recovery, it appears that the low recoveries by the hand-net technique are due to the presence of the catcher and inefficiency of the hand-net technique itself.”45

Consistently, the problem was not to do with the sticky screen. The experiments with mobile attractants showed that the hand net system was a slow and inefficient way of studying mhesvi’s host preference; instead, the experimenters recommended the use of unaccompanied mobile models coated with sticky deposits and with visual stimuli resembling mhuka.46 Conducted principally at Sengwa, the mobile research was mostly limited to a black drum with ten-gallon capacity mounted horizontally fourteen inches above ground on a little perambulator chassis. This model was moved on a straight run at the speed of 1.5 miles per hour using a long handle, with the operator pulling as he walked twenty-three feet ahead. Alternatively, a long rope was fixed to the chassis so that it could be operated by munhu mutema simply pulling while hiding a hundred yards away.

The experiments with the five attractants discussed previously—in mobile mode—were each conducted at the same time daily, for the same number of days, over a distance ranging from 200 to 1,400 yards. The mhesvi visiting the attractant were counted using the naked eye or by telescope, or those recovered with a drop net toward the end of the run or trapped by sticky deposits were counted manually. The sticky recovery model was covered with black sticky deposits, worn by a mufrayi carrying on his back a 3 ft. × 1 ft. screen with khaki sticky deposits.47 The walking man generally depressed the numbers of mhesvi caught compared to unaccompanied models—the mhesvi was repelled, possibly, by visual stimuli.48

The question now was what to do with mhesvi after trapping it: Destroy it? Sterilize and release it? Here, we will limit ourselves to the first option—kill—and defer the second—sterilize and release—to the next chapter.

Insecticide Treatment

One method of killing mhesvi was the treatment of traps with insecticide, but as late as 1982, the biggest challenge remained the fact that most flies visiting such targets could not alight on insecticide-coated target surfaces. The solution was to place a sheet of fine netting close to the target and smear it with DDT wettable powder (WP). With this method, almost 100 percent of all mhesvi that collided with or landed on the netting died. There was another problem: Once heavy rains had fallen, the lethality of the netting was virtually washed away. Efforts to find a roof to protect the deposit failed, so BTTC persevered with a second option.49

The insecticide used in the field experiments was deltamethrin, a suspension concentrate that was moderately resistant to rainwash and which was also an ultraviolet light absorber. These two properties reduced the hydro- and photodegradation of deltamethrin considerably and paved the way for the design of a much cheaper but still efficient target without a roof.50 The environmental pollution effects of pesticides that were hard to break down naturally were not considered. Deltamethrin had been proven to be effective when applied to netting at three-month intervals, and it now replaced the dieldrin wettable powder used previously. Deltamethrin also had two distinct advantages: First, unlike dieldrin, deltamethrin did not leave a white deposit on the netting and thus did not suffer the repellence associated with the former. Second, its deposits were more resistant to rainwash than those of dieldrin.51

The 1982 experiments resulted in the conclusion that pesticide-treated netting was more economical than traps baited with carbon dioxide and acetone. As the assistant director noted in his annual report, he hoped that the ongoing research would “lead to the development of targets that can be suitably cheap and convenient to compete favourably with conventional systems of tsetse control.”52 Armed with this knowledge, BTTC made several changes to its trap and target designs. The trap was the F2, originally designed in a rigid form, but remastered into a portable, folding form for redeployment purposes. To increase catches, various flags were placed next to the trap.53 In 1985, a new cage made of plastic bottles was added to the F2, which killed mhesvi and funneled them into a large polythene bag underneath the trap to prevent clogging when the trap was not emptied frequently.54

Field trials of traps reached a milestone in 1984. First, targets baited with acetone and 1-octen-3-ol and coated with deltamethrin were deployed on the 5 km2 Antelope Island on Lake Kariba at a density of four targets per square kilometer. A rapid decline and virtual extinction of mhesvi populations was achieved.55 Meanwhile, an even larger field trial was under way from March to October 1984 near Chirundu, where 2,850 targets were placed in a 600 km2 bush known as the Rifa Triangle. The target used—later called the R-Type (Rifa Type)—was a 0.8 m wide black panel in two planes equipped with a hooded downwind gauze panel swinging on metal support and wire frame. This was the “first offensive deployment” of targets following successful trials on Antelope Island on Lake Kariba. The mhesvi population declined rapidly, reaching near extinction by January 1985, or a 99.9 percent reduction in nine months.56

Given differences in terrain and in the behavior of different mhesvi therein, tactical and material changes in target systems were inevitable. One change related to the materials used in assembling the target. The results of a trial in mhesvirutondo-infested Rukute Valley were much better than previous results, but still much slower than the rapid decline of mhesvirupani. Two studies had shown similar results (Vale et al. 1986, 1988). The experimenters used the following four standard target types (the dates of the experiments conducted are included):

  1. R-Type: 0.8 m wide black panel in two planes with downwind gauze panel, hooded, swinging on metal support with wire frame (Rifa, 1984)
  2. S-Type: single plane, center black panel 0.7 m, with 2 m × 0.5 m gauze panels, swinging on metal and wire frame (1985–1992)
  3. S-Type: all solid black, 1.7 m × 1 m on metal/wire frame (1992–1993)
  4. S-Type: all solid, center panel black, two side panels blue (1994–1995)57

The first S-Type (Sengwa Type) targets proved to be twice effective as the R-Type. The switch to all solid materials was an attempt to prolong the life of the model’s treated surfaces. The use of blue instead of black was meant to increase the target’s visual attraction and thus reduce the amount of pesticide applied, and locally manufactured cotton cloth was the preferred fabric to lower costs. Reducing photodegradation of the color remained a challenge. The insecticide was applied using three methods: knapsack sprayer in situ (1984–1986), dipping and drying at the base camp (1986–1993), and painting on tables at headquarters (1994–1995). The concentration of pesticide varied from 0.1 percent to 0.52 percent.58

An expert report in 1995 cautioned, however, that “the general concern at the apparent lack of success is in fact based on the unrealistic expectation that eradication should be achieved in one year.”59 So long as experiments still used these four standard target types per square kilometer, each for a different type of mhesvi, it would require a significant outlay of material, transport, and labor, which were never always readily available, thus directly affecting the veracity of the results. The report’s conclusion was rather grim: eliminating mixed populations per square kilometer would take four to five years.60

The attractant preferred, acetone plus octenol, was “as good as any other combination” then in use for mhesvi—and cheaper. Baits containing phenol and octenol were best for mhesvirupani, especially when combined with traps. Used alone in a sachet, octenol tended to evaporate faster, thus requiring constant reevaluation and replenishing.61 A more perfected killing trap remained elusive.

Electric Traps

Using electric current to kill mhesvi was Swynnerton’s idea, based on a request from the governor of Tanganyika in the early 1930s to equip all trains passing through the territory with fly traps. To stun mhesvi, voltages of 2–3 kV were required. Experiments to generate such voltage with motor engine–driven magnets and with combinations of a car radio vibrator and ignition coil were undertaken (Swynnerton 1933).

The use of electric current was designed to shock the chipukanana temporarily or kill it. These experiments were undertaken in 1971 and involved deployment of a large electric netting composed of a cage 1.35 m tall × 1.5 m long × 0.79 m wide and a net of 3.3 m long × 1.5 m high, stretching 0.3 m to 3.6 m from one side of the cage to the other. These two sections of cage and net were deployed in such a way that the maximum cross section would be exposed to prevailing winds. Then, a small black model of an animal was placed in the center of the cage, with olfactory stimulus (ox odor) in some experiments and zero bait in others.62

Two components of these electric capture systems were developed in 1971. One was an electric tray, designed to kill mhesvi alighting on the ground adjacent to bait, flying near the ground, or about to alight. The set was designed so that flies were electrocuted by a horizontal grid of fine electric wires and fell into a tray of netting, from which they were harvested as samples. The electric tray was pulled behind mobile baits.

The second innovation was an electric pen, a large, circular cage of electrified netting to electrocute all mhesvi flying toward a real mhuka (normally a black ox weighing 400 kg), shown in figure 4.6 with three “native assistants” standing with it inside the cage.63 Three field assistants wearing khaki uniforms were “placed in the pen on six afternoons” for two hours and fifteen minutes from 1600 to 1815 hours. Their task was to observe and identify mhesvi colliding with the electric netting as they attempted to reach the ox inside (Vale and Phelps 1974, 2). The black men were at once a material part of the apparatus and observers of inbound mhesvi—a means of doing knowledge production.


Figure 4.6 An electric pen using as bait a jon’osi and three men.

Source: Annual Report 1972, 7. SACEMA/TA.

The experiments with the electric pen also compared captures made using hand nets and the electric pen, with an ox and three black men as bait, and those using an electric pen with different baits: (a) an empty pen, (b) a pen with an ox, (c) a pen with an ox and three men, and (d) a pen with three men. Version c is shown in figure 4.6. The BTTC drew two conclusions: First, catches of female mhesvirutondo and mhesvirupani from one ox plus three men were greater when using the electric pen than with hand nets. Second, catches of male and female mhesvirupani using the electric pen increased when using the ox alone compared to the ox plus three men.64 The conclusion was that models were more effective baits than men and that the presence of people reduced the model’s catches.65

The electrocuting devices were designed to kill or stun mhesvi by passing high-voltage electric current and electrocuting it. The stunning and killing effect of a grid of fine wires running parallel, 0.8 cm apart, and forming grounded and charged poles of high-voltage current impressed experimenters. Five versions of the trap were experimented with at Sengwa (Vale 1972, 11):

  1. Electric surface: To capture tsetse alighting on model animals or on the backs of men, an electrocuting grid was placed 0.8 cm above the bait surface.
  2. Electrified decoys: Tsetse visiting baits often alight on or near tsetse already on the baits, perhaps for a sexual purpose. Decoy tsetse were placed on men or model animals, each mounted on a central charged wire and flanked by grounded wires. Decoys of male and female tsetse and even small rolls of cloth were highly effective. As many tsetse were taken from six decoys on a man’s back as from the totally electrified surface of his back.
  3. Electric tray: Many tsetse alight on the ground near baits. To capture such flies, a 90 × 90 cm tray of fine netting was mounted on small wheels 5 cm above the ground and 2 cm below a horizontal electrocuting grid. Tsetse flying near the ground, and presumably about to alight on it, were electrocuted and fell, retained by the netting. The tray was pulled 23–113 cm behind mobile baits.
  4. Electric net: To capture tsetse flying near bait, a vertical net, 90 × 90 cm, of fine nylon was flanked with an electrified grid 0.8 cm from the netting. The net was mounted on small wheels and pulled 23–113 cm behind mobile baits. The recovery chute at the base of the net was divided into front and rear portions to roughly separate flies caught near or far from the bait.
  5. Electric pen: A large circular cage of electric netting, 3.4 m high by 6 m in diameter, was created to capture tsetse flying towards host animals tethered inside.

The system for these experiments was based on “the tsetse shorting a circuit between fine copper wires running parallel, 0.8 cm apart and charged with several thousand volts.”66 The electrocuted chipukanana would fall onto a sticky tray or be guided by a polythene chute to a nonreturn vessel. The electrocution experiments went even further. The ten-gallon black drum simulating an animal and the 3 ft. × 1 ft. screen that the black man in the experiment wore on his back were both electrified to capture alighting zvipukanana, usually hungry females feeding and “sexually appetitive” males chasing them. Ninety-five percent were captured by this method (Vale and Phelps 1974).67

The electric circuit was designed to work with absolute minimal power requirements (i.e., a small, sealed, two-volt accumulator, rechargeable with a solar cell array or car battery), because the trap was deployed in remote areas. The generator emitted energy only when a mhesvi was trapped. The circuit used an inverter with a step-up transformer and voltage-multiplying rectifier (Rogers and Smith 1977, 155; Vale 1974a, 1974b).

Conclusion: Converging Forms of Knowledge and Practices

The design of traps has been examined as a process of making means and finding ways of controlling the mobilities of mhesvi. Different types of mhesvi, different hosts, and different vegetation types required specific trapping devices; the trap that worked superbly against mhesvirupani, for example, did not work against mhesvirutondo. The trap was at once a control and a research device, via which mhesvi were caught and killed or captured alive for research. Its ease of construction, cost-efficiency, and self-operated killing capacity once set made it an attractive alternative to other methods.

The key factor in trapping was attracting the mobile mhesvi to the trap. As we discussed in chapters 1 and 2, the mobilities of zvipukanana and their predators, what attracted them (smell and vision), and how to manipulate them for the human good were already well-known. This chapter has shown how vachena translated and tapped into this ruzivo and practices based upon it. The appropriation was not a straightforward mobility of ruzivo from vatema to vachena, but the subjection of ruzivo to vachena’s ways of knowing and knowledge. The infrastructure of controlling mhesvi outlined in the previous chapter was brought to bear upon this chipukanana, its blood meals, and the habitats under experiment (not just experience) and forms of peer review acceptable to vachena’s knowledge production practices. For vatema, movement by night was a way of avoiding detection and molestation by mhesvi active during the day. They used darkness as a way of managing mhesvi through avoidance. The Rhodesians inverted these workings of visibility and mobility to encourage, not prevent, mhesvi to bite (bait) cattle, which now became instruments for detecting mhesvi presence. Vatema had deployed visibility to avoid the detection of the mobilities of vanhu and mombe, vachena to enhance it.

Even when unseen, an animal could not escape detection by mhesvi. Olfactory stimuli were particularly strong in njiri, nguruve, and dzoma. Mhuka hiding or moving in dense vegetation led vachena to the conclusion that odor was as good a stimulus as sight. When put to experiment, the two stimuli were found to complement each other well as elements of a trapping system. The active elements in the odor were found to be carbon dioxide, acetone, butanone, and 1-octen-3-ol. The discovery of carbon dioxide as the key ingredient in odor attractants should bring to mind the earlier and vatema’s continuing practices of burning dry dung indoors and underneath the bellies of mombe or smearing their skins with fresh dung to repel mhesvi. I have not found written evidence of direct mobility of that ruzivo into vachena’s experiments.

The same cannot be said of the direct movement of ruzivo rwezvehurimbo (birdlime knowledge) into the production of sticky traps. Here, I have drawn on experience as archive, detailing the tradition of making birdlime from chitatarimbo, mukonde, and other rubber trees in rural Africa. Hurimbo production was a maker space where the ruzivo of turning liquid into sticky substance happened. This is not tinkering or trial and error; this is the application of specific ingredients according to a set formula peer-reviewed communally. The space around chitatarimbo becomes a laboratory for that fleeting moment when the boys are out there cutting and cooking—then they are gone, the tree continuing its life as a plant in the wild, albeit bearing the scars of its domestic role in the upbringing of the boy child. So too stand the twisted or broken branches in the tree where mudziti once stood—the meeting point of vanhu’s means and ways of fulfilling culinary desires, on one hand, and shiri rendezvousing, sojourning, or brooding on the other. Two mobilities, one meeting point: the tree.

What we had made and strategically deployed for shiri, BTTC extended to mhesvi. Here, the mobility of ruzivo rwevatema to the core of white tsetse knowledge is very direct: from the poacher Bvekenya and his mubhemhe to experiments with hurimbo (chitatarimbo specifically) and OCPs to produce an automobile trap. As in previous chapters and sections, we have illustrated the value-adding elements, the redirection and adaptation to mhesvi, and the mobility of ruzivo from overseas. Again, there is no cut from Europe and North America and paste into Southern Rhodesia; mhesvi with its pestiferous mobilities demanded recognition of its specific qualities. By the time the reader reaches the discussion of the addition of electricity to trapping devices, I hope it becomes clear that mhesvi was becoming a mobile site where many forms of (chemical, electrical, mechanical, and biological) knowledge and engineering converged—all because of mhesvi and its pestiferous mobilities.


Additional Information

Related ISBN
MARC Record
Launched on MUSE
Open Access
Back To Top

This website uses cookies to ensure you get the best experience on our website. Without cookies your experience may not be seamless.