The scent gland of worker honey bees is usually called the Nasanov (or Nassanoff) gland, after the Russian anatomist who first described it. That gland is located on the upper surface of the abdomen, between the last and next-to-last segments (see plate 7A in Ribbands 1953 and figure 7 in Wenner 1971a). When a bee raises its abdomen and flexes the last segment down, a fragrance exudes from a pouch opened by that motion.

Fragrant components emanating from that gland include the alcohols geraniol (trans-3,7-Dimethyl-2,6-octadien-1-ol) and nerol (the cis-isomer of geraniol). The oxidation products of those components, citral (two isomers), geranic acid, and nerolic acids, are also present (Boch and Shearer 1962; Shearer and Boch 1966; Pickett et al. 1980, 1981). The mixture has a pleasant floral smell to human beings, as do its individual components.

These fragrances are not unique to bees; they also occur in some plants, such as in the herb "lemon balm" (Melissa officianalis; see Burgett 1980), and in leaves of the exotic lemon-scented gum tree (Eucalyptus citriodora). The principal volatile component of the latter is geraniol (results from an unpublished gas chromatographic analysis).

In field experiments one cannot help but notice the exposure of the Nasanov gland when bees feed at dishes of sugar solution, particularly if scent levels are low (Wells and Wenner 1971). Bees apparently do not expose that gland while they feed on flowers (Free 1968), even though von Frisch and Rosch (1926) had reported earlier that they did.

Researchers have sometimes gone to elaborate lengths to exclude the presence of that odor during bee "language" experiments (e.g., von Frisch 1923). Gould, Henerey, and MacLeod (1970) also took pains to seal off the gland, despite the fact that considerable data had already existed that indicated that Nasanov glands were apparently not a factor in recruitment (e.g., data in von Frisch 1947; Wenner, Wells, and Johnson 1969; see also Wells and Wenner 1971, 1973).

In chapter 6 we noted that von Frisch's 1923 conclusion, that Nasanov gland exudate attracted recruits, was not supported by his later and more extensive experimental results in 1947. However, since that 1947 paper dealt mainly with other important issues (the generation of the dance language hypothesis), we and most other readers paid scant attention to discrepancies in the scent gland data he published therein. Indeed, we were not sensitized to von Frisch's difficulties in this area until our own observations of Nasanov gland exposure failed to correspond with existing interpretation.

The Nasanov gland attractant problem has interesting application to the relationship between paradigm hold and experimental procedure in science. We pursue that topic here.

HOW SHOULD AN ANIMAL BEHAVE TOWARD AN ATTRACTANT?

The "attractiveness" of a pheromonal mixture, as the honey bee Nasanov gland secretion was presumed to be, can only be assessed biologically. Baker, Meyer, and Roelfos defined attraction as "net within-plume displacement toward the [chemical] source" (1981:269; see also our chapter 5 and excursus OS). An example of an attractant that meets this definition is that of methyl eugenol in the case of the Oriental fruit fly, Dacus dorsalis. Steiner reported a series of experimental tests in Hawaii, in which the movement of Oriental fruit flies toward that chemical was well documented. For example, he wrote:
"In one experiment some 1300 flies were attracted a half-mile to a muslin screen that had been treated with methyl eugenol and exposed to daylight for 5 hours. None appeared until after 15 minutes had elapsed, but 30 appeared during the next hour and 67 during the first 100 minutes" (1952:242).

In addition, catches of flies in areas downwind of such traps declined, and times of first arrivals of flies were correlated with the distances of those populations from the traps. Steiner noted: "The area influenced by methyl eugenol is triangular in shape with the base downwind and the width of the base regulated by wind velocity and directional stability" (1952:242).

Steiner's experimental results, a "net within-plume displacement [of Dacus dorasalis] toward the [methyl eugenol] source," clearly fit within the above definition of "attraction."

HOW DO BEES BEHAVE TOWARD NASANOV GLAND SECRETIONS?

Von Frisch was not the first to suggest an attraction of honey bees to Nasanov gland secretions. In 1901 Sladen concluded that the Nasanov gland attracted disoriented swarm bees, as well as bees returning to the entrance of their hive (in Ribbands 1953:172, 173). However, von Frisch may have been the first to suggest the hypothesis that Nasanov gland exudate was responsible for attracting searching bees to feeding dishes. In 1923, for example, he reported that he had obtained ten times as many new arrivals at stations where Nasanov glands were exposed than at stations where glands were sealed.

Those early von Frisch results seemed to verify Sladen's earlier conclusion that Nasanov gland secretions attract bees. Indeed, von Frisch's 1920s results had so increased his confidence in the "truth" of his hypothesis (that Nasanov gland secretions attract recruits to feeding dishes) that he dismissed anomalous data he gathered during the 1940s, at which time he wrote: "there is no doubt about the existence of an attraction exerted by the scent organ, (c.f. v. Frisch, 1923. p. 155 ff) which has also been confirmed in further experiments into which I do not want to go here" (1947:22).

An analysis of his 1940s data may reveal why he had to dismiss those later results if he was to keep the Nasanov gland attraction hypothesis alive. We reproduce here table 6.2 (as table NG.1) from the body of our text for ready reference (from Wenner 1971a:95).

TABLE NG.1. Summary of results from four experiments, showing the number of recruits arriving at a feeding station and at a test station a short lateral distance from it.
	Food in One Direction from Hive		Food in Opposite Direction from Hive
Date	Food Place at 300m	Station 30m from Food Place	Food Place at 250m	Station 25m from Food Place
		Scent Gland Open
Sept. 17	178	257	-	-
Sept. 18	-	-	362	39
		Scent Gland Closed
Sept. 19	190	228	-	-
Sept. 20	-	-	80	25
SOURCE: Adapted from information on pp. 21 and 22 of von Frish 1947.
NOTE: On two days regular forgers had their scent glands open; on the other two days the glands were sealed out.

Compare the results von Frisch obtained in his experiments 1 and 3 (September 17 and 19). Whether the scent gland was open or closed made no difference in the pattern of recruitment at the two stations - at the feeding station compared to a station 30 meters away from that station. Furthermore, the station 30 meters distant from the feeding station received more recruits than the feeding station itself. That result contradicts expectations, both of the Nasanov gland attraction hypothesis and of the dance language hypothesis.

Results from von Frisch's experiments 2 and 4 reveal further discrepancies. On both of those days (September 18 and 20) the results agreed with predictions that one might make on the basis of the dance language hypothesis, but they contradicted results obtained on the other two days (September 17 and 19). Note also that these September 18 and 20 experiments were run with the feeding station in the opposite direction from the hive. A difference in influences of wind could have been responsible, perhaps, for the discrepancies noted (see our chapter 8).

Many subsequent attempts to verify an attraction "function" of the Nasanov gland secretion have yielded equally puzzling results. Woodrow et al. (1965), for example, tested 195 natural and synthetic odorous compounds on bees in a modified olfactometer. Only four non-Nasanov compounds were "weak to moderate" attractants, and a larger number were repellents. The Nasanov constituents themselves, citral, geraniol, and geraniol acetate, were neither.

Waller (1970) reported the results of field tests with Nasanov components sprayed on an alfalfa field. He used citral, geraniol, and anise (a non-Nasanov component) in attempts to attract recruits to that crop. None of these fragrances regularly increased bee populations in experimental alfalfa plots when applied in water, but each of them singly or in mixtures did so when applied in sucrose solutions. Apparently these odors had thereby merely served as marker stimuli when coupled with a food reward in a conditioned response situation (see our chapter 7).

Preference tests, with dishes placed at several feeding stations, as exemplified by experiments run by Free (1962, 1968), provided another experimental approach by which the attractiveness of Nasanov gland chemicals could be assessed. In one such test (1968) nine dishes were set 90 centimeters apart on a lawn near an apiary. Free had placed excised Nasanov glands near three dishes, three other dishes had whatever residual scent recent foragers might have left, and three dishes were clean and empty. "Inspection" visits and landings of "scout" bees were recorded.

In the ten tests run in Free's experiment, 158 inspections occurred at the three Nasanov-containing dishes, and twenty-five landings were recorded. On the other hand, 227 bees inspected the six non-Nasanov dishes, with twenty-four landing thereon.

Earlier experiments by Free (1962), using geraniol-scented dishes, an alternative non-Nasanov scent, and unscented dishes (three of each) yielded similar results. There were 577 inspections or landings on the three geraniol-containing dishes and 772 on the six nongeraniol dishes. Under the verification approach, one can focus on the slightly greater frequency of success at the dishes marked by geraniol or Nasanov glands, as Free did (1962, 1968). For example, Free wrote:
"Experiments are described which show that geraniol, the principal volatile component of honeybee scent-gland secretion, is attractive to foragers, but that it is not nearly as attractive as scent-gland odour itself" (1962:52).

Alternatively, it is conceivable that the experimental results could reflect no more than a difference in odor intensity. If the Nasanov odor were a true attractant, according to the Baker, Mayer, and Roelfos (1981) definition, nearly all 158 approaches to the Nasanov dishes in Free's 1968 study should have resulted in landings, instead of the less than 14 percent that did so. Also, a large majority of the observed bees not landing at the control stations should have moved over into the odor plume emanating from the dishes marked by Nasanov gland odors (see excursus OS).

It is clear that these and similar dish preference tests failed to qualify Nasanov gland secretions as attractant pheromones according to the usual definition.

THE USEFULNESS OF NASANOV SECRETIONS

All of the above is not to say that Nasanov pheromones may not be useful somehow in honey bee studies. Free and co-workers found that traps baited with synthetic Nasanov "lures" were more effective than unscented traps for catching stray honey bees. Also, synthetic lures at in-hive feeders encouraged bees to consume more water or to forage more at pollen substitute sources (Free, Ferguson, and Pickett 1983), or they may facilitate the trapping of stray honey bees (Free, Ferguson, and Simpkins 1984). Useful as Nasanov lures may be, however, preference for that scent over no scent at all does not characterize the glandular secretion as an "attractant."

It is also well documented that dispersal of the Nasanov odors by fanning and scenting workers apparently help disoriented bees to find their hive entrance (Sladen 1901; Ribbands and Speirs 1953). If one shakes bees off a comb onto the ground in front of a hive, workers at the entrance begin fanning and expose their Nasanov glands. Disoriented bees then walk toward the hive entrance. However, nondisoriented bees from adjacent hives are unaffected by the pheromone at the time.

Beekeepers have also observed heavy use of the Nasanov gland during the settling of a newly emerged swarm, a phenomenon reported as early as 1901 by Sladen and later by Witherell (1985:828). According to Morse and Boch (1971), disoriented bees in a swarm may be induced to land by Nasanov gland exposure. Again, nonswarming bees from nearby hives are not attracted to that aggregation.

OUR EXPERIMENTS AND CONCLUSIONS

Our own disillusionment with the established doctrine of Nasanov gland attractance arose during our "crucial" test of the dance language hypothesis (Wenner, Wells, and Johnson 1969; see also our chapter 10). We noted that we had minimal recruitment of new bees on days when we used little or no odor at the feeding station. However, we also noticed heavy exposure of the Nasanov glands by foragers at the dishes whenever odor level was low. On high recruitment days, when we used scent at the stations, there was little Nasanov gland exposure. These observations contradicted what we expected from the Nasanov gland attraction hypothesis.

Eventually we found that we could control Nasanov gland exposure levels to some degree by altering the amount of scent in the food (Wells and Wenner 1971). A similar inverse correlation was obtained in results on experiments with dance frequency; with less odor at the feeding station, dance frequency increased in the hive. The following table (from Wenner 1971a:94) summarizes that information:

TABLE NG.2. Experimental results when using either scented or unscented food.
Station Condition	Trips	Nasanov Use	Number of Dances	Recruits
Scented food	239	19	49	35
Unscented food	242	169	114	4
SOURCE: Wenner 1971a: 94.
NOTE: The less scent in the food, the more frequently bees danced in the hive. Nevertheless, recruit arrivals plummeted, despite a marked increase in use of the Nasanov glands at the feeding station by foragers.

It was as though foragers were attempting to attract recruits when odor was not present. (Note the anthropomorphic and teleological implications here.) However, recruitment rate increased neither with increased dancing nor with increased Nasanov gland exposure. Therefore, Nasanov glands can hardly be considered to produce an "attractant" according to the accepted definition of that term.

Two other anecdotes apply here. In midsummer of one year, when hives had little nectar to forage on, we opened a bottle of geraniol 200 meters upwind from a strong colony. During a three-hour period, only one bee flew close to the bottle. The second observation is that lemon-scented gum trees are common on the University of California (Santa Barbara) campus. A gas chromatographic analysis of their leaves (mentioned above) revealed that their strong odor was largely due to the geraniol component of those leaves. Near those trees, the fallen leaves exude a pungent odor of geraniol, but honey bees are never seen inspecting the area.

After all of our experimentation and review of the literature, we had to conclude: "None of the evidence [we] obtained supports the hypothesis that Nasanov secretion contains an attractant pheromone. . . . The Nasanov scent appears to provide a point of orientation for confused bees, but bees engaged in normal activity are not attracted to it" (Wells and Wenner 1971:208).

REACTION TO OUR FINDINGS

Even though the question of Nasanov gland function may appear to some to be peripheral to the dance language controversy, this episode provides another example of the application of different scientific approaches during the course of research on a specific problem. Von Frisch's initial "exploration" approach yielded evidence in support of an "attraction function" for the Nasanov gland exudate. During subsequent years he continued his verification approach with regard to that attraction hypothesis (as indicated above), and others followed suit.

When von Frisch and others obtained negative results with respect to the Nasanov gland attraction hypothesis during the mid-1940s and later, it was apparently too late for them to recognize the implications of such negative evidence. They had passed the point of commitment (see figure 3.2), moved into a paradigm hold, and dismissed those negative results.

Most subsequent researchers followed the verification approach as well. Any positive bias in the data obtained from experimentation was seized upon as support of the hypothesis; negative results were either dismissed or ignored. Even at this time, the Nasanov gland exudate is still sometimes considered an "attractant" by proponents of that hypothesis, as expressed by Witherell: "The scent from [the Nasanov] gland is used to attract other bees" (1985:827).

That attitude was also exemplified by Seeley's comments: "If the recruit target lacks significant odor ... bees will mark the site with scent from their Nasonov glands (Free 1968, Free and Williams 1970)" (1985:86).

In the tradition of the verification approach to science, neither Witherell nor Seeley made reference to studies that produced data in conflict with that hypothesis. It is noteworthy, however, that the frequency with which Nasanov gland exudate is mentioned as an "attraction" pheromone has decreased in recent years. Careful reading of the above quotation reveals that even Seeley did not claim an attractive "function" for the exudate, and he published little more on the subject than the above brief comment in his book on honey bee ecology.

The "function" of the Nasanov gland thus remains largely a mystery, but it is becoming apparent that the bee research community has been "backing away" from the von Frisch interpretation that emanations from that gland attract searching recruits. Note, however, that publications on this subject do not directly challenge von Frisch's interpretation.

Instead, what we seem to have here is a gradual (subdued) paradigm shift, wherein both positive and negative evidence are simultaneously deemphasized. In addition, it is evident that no one is credited with refuting the former prevailing hypothesis. We thus have here a category apparently unrecognized earlier by either Kuhn or subsequent sociologists of science; paradigm shifts may occur and remain unrecognized as such when issues and/or personalities are relatively unimportant.