Adrian M. Wenner(1)

Accepted March 8, 2001; revised September 20, 2002

Beekeepers could assist growers greatly if they could direct honey bees (Apis mellifera) from their hives to one specific crop or another; that was the goal of a group of Russian bee researchers and of Karl von Frisch in the 1930s and early 1940s. By the simple process of inserting odor into a colony, they could increase visitation to a crop. Ribbands (1953, p. 184) summarized some of the odor-directed results obtained by two Russian researchers; he reported that Kapustin obtained a 24-fold increase in honey bee visits and a doubling of the seed crop, while Gubin had a 19-fold increase in the honey bee population on red clover, with a trebling of the seed crop. Gubin also found that bees successfully trained by feeding scented syrup inside the hive visited vetches, sunflowers, and lucerne.

Earlier, von Frisch had insisted that the "language" or "speech" of bees during recruitment of naive bees to food sources involved only the use of odor and published a well-developed, though not concise, statement that effectively constitutes an "odor-search" hypothesis (von Frisch, 1937; as in Wenner, 1993). Von Frisch (1943) also conducted research similar to that of the Russians, obtained much the same results that they did, and summarized some of his results, as follows (in translation):

In feeding a group of bees at a scented base, their hive mates, alerted by their dances, scour the locality in all directions in search of that odor. Feeding inside the hive in a scent-laden surrounding can be just as effective as feeding outside .... In the case of red clover the visitation can be increased in this way by 22-fold; in the case of potherb by more than 12-fold. Through these measures the intensity of the work of the bees is also increased and their working hours lengthened. (emphasis mine)

Von Frisch also recognized the overriding importance of wind direction when he wrote, "An appreciable degree of directed flight can be obtained . . . by inside feeding in a scented atmosphere . . . weather conditions, among them wind direction, assuming a special role . . ." and noted, "It was an enchanting sight to watch the bees, often in formation flight just above the ground, heading towards the lavender cardboard against a gentle breeze . . . . Their delay and less frequent coming [in one case] was, as a matter of fact, influenced by the wind direction" (emphasis mine).

That promising line of research (directing bees to crops by the use of odor inserted into the hive; while heeding the importance of wind direction) ceased with the advent of von Frisch's (e.g., 1947) dance language hypothesis. At that time he noted that a successful forager, after returning to its hive, executes a "waggle dance" maneuver that contains distance and direction information (very inaccurate, as learned much later) about the food source it has exploited. Eventually, many naive bees (recruits) from that hive find the same food site as visited by the original forager. Von Frisch concluded that the recruit bees "read" the quantitative information in the waggle dance and use that information in their search - though he had no direct evidence that such was the case.

For clarification, one can state the original intent of the von Frisch hypothesis as follows (slightly modified from Wenner [1971, p. 30] and Wenner and Wells [1990, p. 64]; a statement that has apparently remained unchallenged since its first publication three decades ago).


Postulates

1.	A bee successful at exploiting a source of food in the field succeeds in stimulating other bees (recruits) to leave the hive and search for the same source.
2.	The successful forager, while stimulating others to leave the hive, executes a "dance" upon the surface of the comb. That dance maneuver contains quantitative direction and distance information. A human can "read" the dance maneuver and deduce the approximate location of the food source exploited by the forager.
3.	"Recruits" soon arrive at or near the site exploited by the dancing bee they contacted before leaving the hive.

Conclusion

Recruits can use the direction and distance information provided by successful "dancing" bees and fly directly out to the appropriate location.

Thus it was that bee researchers actually had the equivalent of two hypotheses under consideration by the late 1940s: the quite practical but ill-defined odor-search hypothesis of von Frisch (1937, 1943) and of the Russian workers (e.g., Ribbands, 1953), as well as the more exotic dance language hypothesis (e.g., von Frisch, 1947). Unfortunately, virtually all research emphasis after that time became focused on the language notion. For example, in his massive 1967 summary volume, von Frisch did not stress the importance of wind direction during recruitment, a factor that he had considered important earlier. The same held true for my early research on the subject (e.g., Wenner, 1959, 1962, 1964).

Nor could I find any citations or comments about von Frisch's 1937 and 1943 papers in any readily available summaries of his work (e.g., von Frisch, 1950; Lindauer, 1961; Wilson, 1971; Michener, 1974; Seeley, 1985; Free, 1987; Winston, 1987; Gould and Gould, 1988; Crane, 1990; Gary, 1992; Seeley, 1995). Not until the work of Friesen (1973) did the importance of wind resurface, research that remained largely ignored until only recently (e.g., Wenner and Wells, 1990; Wenner, 1998b-d).

We have now had the honey bee dance language hypothesis (e.g., von Frisch, 1947, 1967; Wenner, 1971) for half a century. In addition, the controversy that has swirled about that hypothesis (i.e., whether recruited bees actually use the dance maneuver information) has existed for a third of a century. The remarkable persistence of both the language hypothesis and the controversy provides many lessons about the nature of scientific inquiry (e.g., Wells and Wenner, 1973; Rosin, 1980; Veldink, 1989; Wenner and Wells, 1990; Kak, 1991; Vadas, 1994; Wenner, 1997). For example, Veldink (1989, p. 175) wrote, "If in a case of pure science, a theory can survive for a dozen years, at least, despite data to the contrary, what are the implications for controversies which have life-or-death consequences?"

Unfortunately, all too often biologists who study behavior disdain lessons provided by scholars outside their own field - for example, input from those who study the philosophy, sociology, and psychology of science. Instead, students usually gain their entry into the field by way of mentorship in one graduate program or other. The quality of their training may thus rest upon a rather limited exposure to the vast amount of information and insight available about scientific process. While some mentors of graduate students provide excellent exposure to a wide spectrum of thoughts about scientific inquiry, most merely strive to help their graduate students mesh into one particular "thought collective" or another, as emphasized by Ludwik Fleck in 1935 (Fleck, 1979, p. 41; see also Wenner, 1997).

Participants can resolve a controversy but must first understand its basis, as in the words of Bruno Latour (1987, p. 62): "We have to understand first how many elements can be brought to bear on a controversy; once this is understood, the other problems will be easier to solve." The treatment that follows examines the bee language controversy from that analytical perspective. I give only highlights here, since space prohibits a more comprehensive treatment - such as given by Wenner and Wells (1990), Wenner et al. (1991), and Wenner (1998a).

Nor does space permit a review of the network of publications that deals with the waggle dance maneuver and its analysis, a topic covered in detail by Dyer (2002).


THE APPEAL OF THE LANGUAGE HYPOTHESIS

In 1946 von Frisch seemed to have tipped the balance away from odor search and toward the notion of language. While many still feel that he "discovered" their language or that he "proved" that bees have a language, he instead only proposed a "language" explanation, as in his words (von Frisch, 1947, p. 5): "Today, after two years of experimenting, I have come to realise that these wonderful beings can, in a manner hitherto undreamt of, give each other exact data about the source of food."

Ten years earlier, Julien Francon had reached the same conclusion and had written (see Wenner and Wells, 1990, p. 55), "The bees communicate with each other, and are even capable of transmitting instructions with a precision that is sometimes astounding" (emphasis Francon's). Whereas Francon had based conclusions upon insufficient evidence, von Frisch had conducted easily repeatable experiments that yielded quantitative results in support of his hypothesis.

A half-century ago, however, "testing" a hypothesis in behavioral studies meant little more than gaining confirmatory evidence (e.g., Lakatos, 1970, p. 187). Even so, von Frisch (1947, p. 11) himself entertained the idea of a true test of the language notion when he wrote,

. . . The observation of the different conduct in the hive of those bees foraging near and far had brought confirmation with unexpected clarity. It did not seem advisable to check this by following up the behaviour of the newcomers. We should hardly have found out anything more than we knew already. (emphasis mine)

However, by not "following up the behavior of the newcomers," von Frisch missed an opportunity. He could have found that recruited bees do not "fly directly out" by use of the crude distance and direction information that exists in the dance maneuver (see below). In fact, no recruits would succeed without the use of an odor cue, as he himself had insisted upon earlier (von Frisch, 1937; see also below and Wenner, 1993).

Not uncommonly scientists veer away from pursuing a research path that might yield unfavorable evidence. Claude Bernard ([1865] 1957, p. 40) had cautioned against such a tack nearly a hundred years earlier: ". . . when we have put forward an idea or a theory in science, our object must not be to preserve it by seeking everything that may support it and setting aside everything that may weaken it."

Instead of testing his hypothesis, then, von Frisch continued to design and conduct experiments that yielded results in agreement with his conclusion that searching bees could use the distance and direction information present in the waggle dance of foragers. He later clarified his conclusions:

We see that the majority of searching bees fanning out, moved within an angle deviating not more than 15 degrees each to the left and to the right from the direction leading towards the feeding place. (von Frisch, 1948, p. 10). For almost two decades my colleagues and I have been studying . . . the 'language' of the bees: the dancing movements by which forager bees direct their hivemates, with great precision, to a source of food. (von Frisch, 1962, p. 78) . . . These creatures can inform their comrades of a goal that is of importance for their colony and can describe its location so exactly that the hivemates find it independently in flight, without being led there, by the most direct route - even at a distance of kilometers. (von Frisch; 1967, p. 524).

More recently Seeley (1995, p. 36) slightly weakened but otherwise concurred with von Frisch's original intent.

When a worker bee discovers a rich source of pollen or nectar, she is able to recruit nestmates to it and thereby strengthen her colony's exploitation of this desirable feeding site. The principal mechanism of this recruitment communication is the waggle dance, a unique behavior in which a bee, deep inside her colony's hive, performs a miniaturized re-enactment of her recent journey to a patch of flowers. Bees following the dance learn the distance to the patch, the direction it lies in, and the odor of the flowers, and can translate this information into a flight to the specified patch. Thus, a waggle dance is a truly symbolic message, one which is separated in time and space from both the actions on which it is based and the behaviors it will guide.

That is, von Frisch had indicated what one could expect recruit bees to do under his (unstated, as such) hypothesis - after they had attended the waggle dance and left their colony. In that sense, von Frisch had met a criterion specified by Alan Chalmers (1978, p. 45): "The more precisely a theory is formulated the more falsifiable it becomes." By such a statement Chalmers followed the lead of the eminent philosopher of science, Karl Popper (1957), who wrote (see also Wenner and Wells, 1990, p. 22): "One can sum up [all of the above] by saying that falsifiability, or refutability, is a criterion of the scientific status of a theory" (emphasis Popper's).


EROSION OF THE DANCE LANGUAGE HYPOTHESIS

According to Popper (1957; summarized by Wenner and Wells, 1990, p. 22), "It is easy to obtain confirmations, or verifications, for nearly every theory - if we look for confirmations," and "Every genuine test of a theory is an attempt to falsify it, or to refute it" (emphasis Popper's). Unfortunately, for the first two decades of its existence, researchers only sought further confirmation and failed really to test the dance language hypothesis. By the mid-1960s the notion of "discovery" or "proof" of bee language had gained so much appeal that no supporters would execute such a test - bee "language" had become "fact" in the minds of most people in the scientific community (in that connection, see Steinbeck, [1941] 1962, p. 180; cited by Wenner and Wells, 1990, p. 110).

While conducting doctoral research at the University of Michigan in the late 1950s, I discovered a highly structured sound that foragers produce during their dance maneuver, a sound pattern from which one can estimate the distance that a forager had flown from hive to feeding place (Wenner, 1959, 1962).

However, a careful analysis revealed that a great deal of variation exists in the sound patterns and in other elements of the dance maneuver. The supposed accuracy of "use" of dance maneuver information that von Frisch obtained in his "step" and "fan" experiments, for example, exceeded the accuracy of information contained in the dance itself (e.g., Wenner, 1962; Towne and Gould, 1988; Weidenmuller and Seeley, 1999; see also below). Apparently, von Frisch's experimental designs (an array of test stations) funneled searching recruits toward the center of all sites in the array (e.g., Wenner and Wells, 1990, pp. 331-338) - that artifactual condensation of recruit search efforts also occurred in experiments run later by others (e.g., Gould, 1976; Towne and Gould, 1988; see below).

Research done with eyes wide open provides curious twists. While trying to construct an imitation dancing bee that might "send" real bees to discrete food source locations, we stumbled onto the disconcerting notion that bees learn quickly (the conditioned response phenomenon, as with Pavlov and his salivating dogs). That realization perhaps should have come as no surprise - but for the fact that by then bee researchers and others had considered bee language an "instinctual signaling system" during recruitment to crops that would not involve learning. Von Frisch (1962, p. 78) expressed that attitude as follows: "The brain of a bee is the size of a grass seed and is not made for thinking. The actions of bees are mainly governed by instinct."

Despite strong resistance by anonymous reviewers, we managed to publish the results of our experiments on learning in honey bees (summarized by Wenner and Wells, 1990, pp. 111-128). However, our finding had far more serious implications - the experiments described in von Frisch's classic 1950 Cornell University Press book dealt only with the re-recruitment of experienced bees, a success that could be explained solely by their reliance on odor and conditioned response. If bees had a language, such an ability would then apply only to the flight out of the hive by inexperienced bees.

A 1966 event at the Salk Institute in La Jolla, California (Wenner and Wells, 1990, pp. 353-361), led to a series of experiments, the results of which appeared later in the journal Science. The first set of experiments relied on a rigorous double control design, in which inexperienced bees would either use information they had obtained from the waggle dance or search for the odor of the food source exploited by experienced foragers (Wenner and Wells, 1990, pp. 151-172). During those experiments, foragers from one hive visited only one site, while foragers from another hive visited that site and three others.

The results were unambiguous; successful bees from both hives had ended up in a similar distribution pattern at all four test stations according to the geometrical placement of those stations (e.g., Wenner, 1998a, Fig. 4). Searching recruits had apparently relied on the odor of the target sources and had ignored any physical information they might have obtained from the waggle dance maneuver before leaving either hive.

A second set of experiments (also published in the journal Science) involved a more rigorous strong inference design (Chamberlin, [1890] 1965; Platt, 1964). Again, successful searching bees arrived at stations that had the appropriate odor but failed to arrive at stations supposedly indicated by dancing foragers in the hive (Wenner and Wells, 1990, pp. 173-186). Also telling: Very few searching bees arrived at stations that had no odor, even though regular foragers visited such stations, and even though they had exposed their scent glands (Nasanov glands) at the feeding places (see below).

Although we did not realize it until decades later, we had stumbled onto the significance of von Frisch's 1937 (p. 35) conclusion in the matter (see also Wenner, 1993):

It is clear from a long series of experiments that after the commencement of the dances the bees first seek in the neighborhood, and then go farther away, and finally search the whole flying district . . .. So the language of bees seemed to be very simple . . . . In performing this experiment I succeeded with all kinds of flowers with the exception of flowers without any scent. And so it is not difficult to find out the manner of communication. When the collecting bee alights on the scented flowers to suck up the food, the scent of the flower is taken up by its body-surface hairs, and when it dances after homing, the interested bees following the movements of the dancer bee and holding their antennae against its body, perceive the specific scent on its body and know what kind of scent must be sought to find the good feeding-place announced by the dancing bee. That this view is correct can be proved easily.

One can also stand at a test station during recruitment experiments, look downwind, and see that recruits arrive only from that downwind direction (as von Frisch phrased it, "against a gentle breeze"). Viewing further downwind with binoculars reveals that the recruits exhibit the same classic zigzag odor-search behavior as exhibited by other insects that home in on an odor source (e.g., Carde, 1984; Wenner and Wells, 1990, pp. 320-330). So far, though, we have not managed to get language advocates to execute this simple exercise.

Unfortunately, von Frisch's earlier and unclarified odor-search hypothesis had become suppressed and/or lost after he proposed the dance language hypothesis. (All too often, the exotic explanation becomes favored in animal behavior experiments.) In fact, von Frisch had apparently forgotten much of his earlier stance when he noticed that some recruits had succeeded without having attended a dancing bee: he wrote (von Frisch, 1947, p. 13), "It follows further that a communication can be transmitted from the returning bee to other bees by touch alone, without the necessity for any dance" (note his failure to mention odor or conditioned response).

For the scientific community, nevertheless, two competing hypotheses existed by then, as indicated in the Introduction. One, the odor-search hypothesis, would have bees behaving in a manner consistent with the behavior of other insects that search for the odor of food sources (or that search for emitted pheromones). The other, the language hypothesis, would have bees performing at a far higher level of complexity (e.g., Rosin, 1980).

Rosin emphasized the disparity between insect-like and human-like possibilities in the case of honey bee recruitment and insisted upon application of the principle of parsimony (Occam's razor) or Morgan's canon (see Rosin, 1980, p. 463): "Morgan stressed that his Canon only prohibits imputing to any animal a higher psychic faculty than is necessitated by the evidence at hand." In other words, one should not credit bees with a "language" if a more simple odor-search possibility suffices. The results of subsequent research, in fact, have reinforced von Frisch's (e.g., 1937, 1943) earlier and more simple odor-search explanation.

Various means exist by which one could test the dance language hypothesis. In von Frisch's own words (as above), one such avenue would be "to check this by following up the behaviour of the newcomers." In short, Do searching bees "fly directly out" from their hive to (and only to) the target source?

In experiments conducted in the late 1960s and early 1970s, Friesen (1973, Fig. 15 and Table III; see also Wenner 1998b-d) found that newly recruited bees required far too much time after leaving the hive and before arriving at a feeding station to have "flown directly out," as would have been the case if they had used the distance and direction information contained in the dance maneuver of foragers - as implied in the original von Frisch language hypothesis.

Somewhat earlier, Gould et al. (1970) reported that ". . . delays between the recruits' dance attendance and arrival at a feeding station were distributed almost uniformly from <1 minute to 9 minutes," compared to a flight time of only 16 to 18 s for foragers (see Wenner and Wells, 1990, pp. 302-308). In fact, one recruit in their experiments searched for 75 min before reaching a target station.

The poor success ratio for searching recruits in the Gould et al. experiment mirrored the lengthly search times they had recorded; 277 bees left the hive after having attended 155 observed dances. Of those 277 recruits, only 37 found either of the two stations located only 120 m from the hive. Twenty-five of them ended up at a station in the "correct" direction, but 12 of them actually ended up at a station in the opposite direction, one that had not been "indicated" in the dance maneuver. These serious discrepancies fazed neither those researchers nor the editor of the journal Science (see Wenner and Wells, 1990, pp. 274-284).

Esch and Bastian (1970) published results with similar implications, as summarized by Wenner and Wells (1990, p. 217). Only 14 of the 70 bees that had attended forager dances and left the hive found the target station. Ten of those 14 recruits required between two and nine exploratory flights (after repeated contacts with forager dances between flights) before they located the station; thus, only 4 of the 14 successful recruits located the food on the first flight. The average time for arrival by successful recruits was 8 min, compared with the half-minute needed for a "beeline" flight between hive and station.

Gould (1976, p. 228) later recognized the problems occasioned by the long search times of recruited bees, agreed that recruited bees took too long to reach a target station, and admitted that ". . . the statement that recruits 'fly rapidly and with certainty' (von Frisch, 1967) is subject to doubt."

If von Frisch's 1937 odor-search hypothesis had remained prominent in the literature, the accumulated negative evidence with respect to the com-plex dance language hypothesis might have had an impact later on. Instead, the deep entrenchment of the more exotic language hypothesis resulted in an almost-predictable response by the scientific community, as phrased by Lindegren (1966, p. 6): "The flaws of a theory never lead to its rejection . . . . Scientists tolerate theories that can easily be demonstrated to be inadequate."


RESCUE ATTEMPTS

Instead of heeding the problems occasioned with accumulated negative results, various researchers attempted to reconfirm the language hypothesis experimentally (e.g., Wenner and Wells, 1990, pp. 207-228). Most of them acknowledged discrepancies between their results and the expected results and then provided a set of qualifications and "apologies" for those anomalies. For example, Gould (1976, p. 239) concluded, "von Frisch's controls do not exclude the possibility of olfactory recruitment alone . . . ."

Fleck ([1935] 1979, pp. 30, 31) recognized the weakness of attempts to reconcile anomalies that arise with respect to entrenched theory: "The very persistence with which observations contradicting a view are 'explained' and smoothed over by concillators is most instructive. Such effort demonstrates that the aim is logical conformity within a system at any cost, and shows how logic can be interpreted in practice."

One should also note that, to date, apparently no one has been able to repeat Gould's (e.g., 1976) "misdirection" experiments; hence, we lack a prime requisite in scientific inquiry - repeatability, in that case. Furthermore, that research of Gould has come under severe criticism by Ohtani, Rosin, and others (summarized by Wenner and Wells, 1990, pp. 231-254; see also pp. 274-284), as has the use of the array design in such experiments (Wenner and Wells, 1990, pp. 331-338).

Eventually, the accumulated anomalies seeped into the collective scientific consciousness, at which time some qualifications concerning the efficacy of the dance language hypothesis began to appear in the literature (e.g., Winston, 1987, pp. 157, 160).

Towne and Gould (1988) studied the "spatial precision" (error in direction and distance information) of the dance maneuver at different distances. According to their measurements, at a distance of 500 m the standard deviation in direction information in a forager's dance maneuver (with an average divergence angle of 15º) would be about 16º (as in their Fig. 12). In an earlier study, Wenner (1962, Table 1), had found that the error (standard deviation) in distance information would be about 100 m at about that distance. Together, those two values would thus "describe" an area of more than 30,000 m2 at 500 m from the hive, hardly the precision von Frisch (e.g., 1948, p. 10, 1962, p. 78, 1967, p. 524) had claimed earlier. Of course, a great deal of error in searching behavior would also hold true for searching bees - if they could actually use such information.

Towne and Gould (1988) then employed the flawed array design (see Wenner and Wells, 1990, pp. 331-338), as used earlier by Gould (e.g., 1976) and others, but still found a wide divergence in recruit search areas (their Fig. 13). Not surprisingly (given their use of the flawed experimental design), the <21,000-m2 search area of recruits that they estimated for the 500-m distance was less than the error present in the dance maneuver itself.

Instead of heeding the negative implications of their results, Towne and Gould (1988, p. 152) molded their findings into a new conception of the language hypothesis, one that fit into a "tuned-error" hypothesis: ". . . If a patch of food consists of a collection of small 'packages' scattered over some area, there is a single, nonzero value for the scatter of recruits that yields the optimum overall foraging performance, and the more scattered the food, the greater the optimal scatter."

In other words, Towne and Gould rationalized dance maneuver error into an "advantage." Thus, whereas von Frisch used one ad hoc device (that recruits average distance and direction information from several dances to find the target source), Towne and Gould had used another ad hoc device (error is good) in an attempt to dismiss the problem of too much error in the dance maneuver and in the supposed "use" of that information. Earlier, Free (1987, p. 120) had employed a similar rhetorical device: "Bees recruited to an attractive natural food source do not follow the dance directions very precisely . . . nor is it desirable that they should do so."

Weidenmuller and Seeley (1999, p. 198) voiced support for that rationalization when they reported, ". . . we believe that the evidence we present provides strong support for the tuned-error hypothesis of Towne and Gould (1988)." Neither group, though, apparently realized that their conclusions stood in stark contrast to von Frisch's original intent and to Seeley's (1995, p. 36; see above) earlier endorsement of that intent.

Another ad hoc modification frequently encountered in the bee language debate is the proposal that bees sometimes use "their language" and sometimes use odor, as Gould and co-workers suggested when they wrote a disclaimer (as given by Gould, 1976, pp. 239-240): "Simply demonstrating that olfactory cues are sufficient in a particular situation does not mean that the dance language is not used under other conditions" (emphasis mine; note an assumption there of bee "language" as "fact".

A new twist to the controversy stems from yet another attempt to accommodate both "use of language" and use of odor (Donovan, 2000, p. 7).


Two Profitable Uses for the Same Information

A bee following a dancing forager has two possible ways of using all the information in the dance that could maximize its foraging success and thus the competitive foraging success of the hive:
1.	follow the distance and direction information indicated by the dancing bee (and when at close range the odour) to reach the location of the new food that the dancing bee came from, or
2.	use the distance and direction information of the new food location to avoid that location and to set out in other directions to find a new, unexploited location of the same new food, using the odour information imparted by the dancing bee to find plumes of similar odour" (emphasis Donovan's).

Again, we see another ad hoc modification employed to rescue the dance language hypothesis (use information to avoid a food source). Such evasive action, however, does not mesh with the advice given by Karl Popper (1957; as cited by Wenner and Wells, 1990, p. 22): ". . . Every 'good' scientific theory is one which forbids certain things to happen; the more a theory forbids, the better it is . . . . A theory which is not refutable by any conceivable event is non-scientific." Lakatos (1970, p. 96) later stressed that point more strongly: "Scientific honesty . . . consists of specifying, in advance, an experiment such that, if the result contradicts [a] theory, the theory has to be given up."

Such claims and excuses may satisfy those who wish to believe in bee language, but the reasoning has little scientific merit. One has then lost an objective ability to predict the outcome of an experiment in any prescribed set of circumstances.

Employment of a scented mechanical model of a dancing bee (e.g., Michelsen et al., 1989) furnished some sparse results in support of the language hypothesis. Their experiments on the supposed use of distance and direction information by searching bees, as recruited by a scented "robot" bee, yielded results that also actually agreed with a random odor-search model for a food source (e.g., Wenner et al., 1991; Wenner, 1998a) but not with what one would predict on the basis of the original language hypothesis. For example, the vast majority of recruits did not arrive at the specific distances supposedly indicated by the dance maneuver (Wenner et al., 1991; Wenner, 1998a).

Later experiments by Michelsen and co-workers (1992) again yielded some supportive evidence but suffered from flaws in experimental design. For one, observers tallied but did not catch bees that inspected test dishes; they wrote (Michelsen et al., 1992, p. 144), "It is possible, therefore, that some bees may have made two or more approaches at the same or at different locations." In fact, during the 3 h that each experiment was run, some bees could have been counted several times.

In addition, the Michelsen et al. experiments do not appear to have been run "blind." That is, observers at the test stations could well have known where the senior investigators expected recruits to arrive, according to the favored hypothesis (see Michelsen, 1993, p. 140). Nor could the observers know that any bees they saw in the area had, in fact, come from the experimental hive and not from other hives in the area.

Dyer (2002, p. 928) summarized that research, as follows:

[The Michelsen, et al. experiments would indicate that] the production of airborne sound is necessary for a mechanical model bee to recruit bees to feeding places in the environment (68). Here again the possibility exists that the sound merely helps followers to stay oriented to the dancer but is not the channel through which spatial information flows. Furthermore, the recruitment efficiency of the model bee is low, suggesting that something beyond the presence of sounds and the correct pattern of body movement is needed for effective communication.

Finally, we already know that an odor stimulus by itself, under the proper conditions (as covered in the Introduction), can result in bees leaving their hive and searching for that odor (e.g., von Frisch, 1943; Ribbands, 1953, ch ap. 23; Hill et al., 1997, Expt 2; O'Dea, 2000).

Swarm movement and attempts to pollinate crops deserve mention at this point. When a swarm is about to move from near its parent colony to a new site, scout bees execute dance maneuvers on the surface of the swarm cluster. Any wide scatter in the waggle dance information also undermines the notion that bee swarms relocate by use of that information. Weidemuller and Seeley (1999) reported that those dance maneuvers show less variation than the ones foragers execute after visiting food sources. Even so, swarms move through the air and end up at a new location with far more accuracy (a single point in the landscape) than expected on the basis of information contained in the dance maneuver. Instead, extensive research on the highly effective use of artificial chemical lures (synthetic Nasanov gland secretion) placed in swarm hives (e.g., Schmidt, 1994) implicates odor as a major factor, perhaps the only factor, in swarm relocation (see Wenner, 1992).

In addition, and contrary to some reports (e.g., von Frisch, 1947; Free, 1987), the Nasanov gland secretion apparently does not attract searching bees to food sources (e.g., Waller, 1970; Wells and Wenner, 1971; Wenner and Wells, 1990, pp. 312-319; Wells et al., 1993; Winston and Slessor, 1993, p. 19). Instead, that secretion apparently both coalesces groups of disoriented bees and guides bees to a new location during swarm movement (e.g., Wenner, 1992). In that connection, Schmidt (1999, p. 2055) concluded, "Nasonov secretion meets all the criteria necessary to be a pheromone - it is released by individuals to attract other individuals of the species to a specific location; the receivers respond by being attracted to the pheromone source, and the pheromonal response apparently is not elicited by other known odors or secretions."

Failure to abandon a hypothesis that has not survived testing fits into what I consider the Humpty Dumpty syndrome (as expressed in the familiar nursery rhyme) - " . . . And all the King's horses and all the King's men could not put Humpty Dumpty back together again" - which helps explain why the bee language controversy has persisted for so long. Despite the language hypothesis being "broken" (i.e., having failed critical experimental tests), supporters seem unable to resist the temptation to patch the pieces together to make it seem whole again.

All of the above attempts to rescue the language hypothesis, coupled with the twin influences of teleology and anthropomorphism (e.g., Wenner and Wells, 1990, pp. 362-366), have impeded, not enhanced, progress in our understanding of foraging and recruitment behavior in honey bees.


THE WAGGLE DANCE: A SYMPTOM OR A SIGNAL?

Does a dancing bee "intend" to send its hivemates out to a profitable food source or likely homesite? Consider an analogy. In the same vein, one could consider that male crickets while chirping actually "broadcast" a rather complex "message." A knowledgeable researcher can listen to the chirp and recognize the species. Need one then conclude that the cricket thereby intentionally communicates its species identification to other crickets, to other animals, or to us?

Given enough study, a researcher might also infer from the chirp pattern whether that cricket is alone, whether a female is nearby, or whether two males are engaged in a territorial encounter. Once knowing the identity of a cricket species, additional study would permit one to gain an estimate of the temperature at the site of the cricket chirping.

Thus, an examination of cricket chirping can provide several "symptoms" about what goes on in nature, as in the above example. However, one need not conclude that a male cricket engages in chirping behavior in order to inform female crickets about the temperature near where he chirps. No such intent (purpose) is implied by the chirping. We could also ponder the question, "Why do crickets chirp?" and could as easily conclude that the sound "is meant to" give us pleasure during warm summer nights. In essence, a nonadaptive feature need not be eliminated by natural selection (e.g., Gould and Vrba, 1982).

Consider the honey bee waggle dance maneuver in the same light. A forager returns to the hive and executes a waggle dance. By examining that maneuver we can gain information about what that bee experienced after it left the hive on its way back out to the food source. The angle of the straight-run portion of the maneuver provides a rough estimate of the direction it flew from the hive (e.g., von Frisch, 1947); the time spent producing sound during that straight run gives us an approximation of the time spent on that outward flight (e.g., Wenner, 1962).

Those facts by themselves represent only a network of symptoms. As Seeley (1995, p. 36) wrote, the dance is ". . . a unique behavior in which a bee, deep inside her colony's hive, performs a miniaturized re-enactment of her recent journey to a patch of flowers." That maneuver is thus only symptomatic of the forager's experience. By themselves, the existence of the waggle dance and the information contained in that maneuver are not evidence of a deliberate communication act.

Recognizing the existence of a symptom represents only the first stage of a scientific investigation. One must then frame scientifically testable hypotheses and accept whatever results emerge during testing of those hypotheses. The original and very scientific notion of "language" that von Frisch proposed provided the appropriate beginning for further research.

After that time, however, the hypothesis failed critical tests and became further weakened by the emergence of additional adverse experimental results (e.g., the length of time recruits take to find the same site and the small percentage of searching bees that find the same food source visited by foragers). Researchers then attempted to rescue the hypothesis with ad hoc qualifications, so much so that we now have many vague dance language hypotheses in print - most of them no longer testable scientifically.

Unfortunately, an undue focus on the question, "Why do bees dance?" has become a major stumbling factor in the bee language debate. The teleological approach, so popular in behavioral studies these days, would dictate that such a behavior must have a function. A conservative approach indicates otherwise - not every action must have a function (e.g., Wenner and Wells, 1990, pp. 362-366). As far back as 1865 Claude Bernard ([1865] 1957, p. 80) warned against such a teleological attitude: "The nature of our mind leads us to seek the essence or the why of things . . . experience soon teaches us that we cannot get beyond the how, i.e., beyond the immediate cause or the necessary conditions of phenomena" (emphasis Bernard's)

The eminent animal behaviorist Jack Hailman (1977, p. 187) echoed that sentiment when he wrote, "It is irrelevant whether the teleology is naively Aristotelian or framed in Darwinian language - it is still incorrect to 'see' communicative design apparent in [bee] dancing."

Teleological thinking most often goes hand in hand with an anthropomorphic attitude, the belief that a specific behavior must have a function that meshes with what we humans might perceive as the most "reasonable" explanation for such a correlation.


WHAT NEXT?

During the last few decades, the "why" question in this episode has gradually become replaced by the more scientific "how" - as even dance language advocates have obtained results (mostly inadvertently) relevant to the point first raised by von Frisch: "It did not seem advisable to check this by following up the behaviour of the newcomers." However, in 1937 and 1943 von Frisch had already published some clear statements on the "behavior of the newcomers," as outlined in the Introduction (see also Wenner et al., 1991; Wenner, 1993), statements that did not mesh with his later language hypothesis.

If the two extant hypotheses (odor-search and "language" use) had both been kept under consideration this past half-century, the question, "How do recruits find the food source once they have left the hive after attending a dance maneuver?" could have led to fruitful research on the foraging patterns of bee colonies, information that would have proven very useful for those interested in improving crop pollination.

In retrospect, research by Friesen (1973) essentially constituted a continuation of von Frisch's 1930s and early 1940s work on the importance of odor and wind during recruitment to crops (see summary by Wenner, 1998b-d). Unfortunately, bee language advocates ignored Friesen's essential extension of von Frisch's pioneering work. Instead, during the next quarter-century, millions of dollars continued to be spent on relatively fruitless examination and reexamination of the dance maneuver (the symptom), as well as efforts to "prove" that bees have a language, after all (despite considerable accumulated negative evidence).

In fact, the repeated attempts to "prove" the existence of bee language in itself constitutes an acknowledgment that previous attempts had failed. In those cases of reaffirmation attempts, only when advocates had gained new (and to them convincing) confirming evidence did they admit that earlier "proofs" had not sufficed (e.g., Gould, 1976).

Bernard ([1865] 1957, p. 39) addressed that approach as well: "If men discuss and experiment . . . to prove a preconceived idea in spite of everything, they no longer have freedom of mind, and they no longer search for truth." See also Karl Popper's (1957; cited by Wenner and Wells, 1990, p. 22) comment on that point.

The words of Nobel laureate Peter Medawar (1981, p. 73) also ring true with respect to this controversy: "It is a common failing - and one that I have myself suffered from - to fall in love with a hypothesis and to be unwilling to take no for an answer. A love affair with a pet hypothesis can waste years of precious time. There is often no finally decisive yes, though quite often there can be a decisive 'no.'" Hopefully, the scientific community will eventually realize that the bee language controversy is one not of evidence but of how one views the available evidence (e.g., Veldink, 1989).

In the meantime, we now no longer have a concise language hypothesis as initially envisioned by von Frisch (above). Instead, we have many vague versions that no longer have predictive value. We can thus ask, "How much more time will pass before bee researchers begin to study the entire system of colony foraging behavior (e.g., Wenner, 1998b-d), instead of focusing so much on mere symptoms?" and "Will language advocates now broaden their perspective, pursue the lead of von Frisch (1943) and Friesen (1973), and, finally, investigate the role of wind in colony foraging patterns?"


ACKNOWLEDGMENTS

Many thanks go to the scores of individuals in various fields who have provided intellectual support in an otherwise hostile environment during these past several decades and to Ruth Rosin for exceptional persistence under the same conditions. Thanks go also to those ardent and vocal bee language advocates who have helped sharpen the issues at stake. I thank John Richards, Justin Schmidt, Robbin Thorp, Patrick Wells, and Dieter Wilkens for helpful comments on the manuscript. Special thanks go to Barry Birkey, who has provided ready access to many of our publications at http://www.beesource.com/pov/wenner/index.htm.


REFERENCES

Bernard, C. ([1865] 1957). An Introduction to the Study of Experimental Medicine, Dover, New York.

Carde, R. T. (1984). Chemo-orientation in flying insects. In Bell, W. J., and Carde, R. T. (eds), Chemical Ecology of Insects, Chapman and Hall, New York, pp. 111-124.

Chalmers, A. F. ([1976] 1978). What Is This Thing Called Science? Open University Press, Milton Keynes, UK.

Chamberlin, T. C. ([1890] 1965). The method of multiple working hypotheses. Science 148: 754-759.

Crane, E. (1990). Bees and Beekeeping: Science, Practice, and World Resources, Heinemann Newnes, Oxford.

Donovan, B. J. (2000). A comprehensive honey bee dance and odour "language" hypothesis. Bee World 81: 5-10.

Dyer, F. C. (2002). The biology of the dance language. Annu. Rev. Entomol. 47: 917-949.

Esch, H., and Bastian, J. A. (1970). How do newly recruited honey bees approach a food site? Z. Vergl. Physiol. 68: 175-181.

Fleck, L. ([1935] 1979). Genesis and Development of a Scientific Fact, University of Chicago Press, Chicago.

Free, J. B. (1987). Pheromones of Social Bees, Cornell University Press, Ithaca, NY.

Friesen, L. (1973). The search dynamics of recruited honey bees, Apis mellifera liguistica Spinola. Biol. Bull. 144: 107-131.

Gary, N. E. (1992). Activities and behavior of honey bees. In Graham. J. M (ed.), The Hive and the Honey Bee, Dadant and Sons, Hamilton, IL, pp. 269-372.

Gould, J. L. (1976). The dance-language controversy. Q. Rev. Biol. 51: 211-244.

Gould, J. L., and Gould, C. G. (1988). The Honey Bee, Scientific American Library, New York.

Gould, J. L., Henerey, M., and MacLeod, M. C. (1970). Communication of direction by the honey bee. Science 169: 544-554.

Gould, S. J., and Vrba, E. S. (1982). Exaptation - A missing term in the science of form. Palebiology 8: 4-15.

Hailman, J. P. (1977). Bee dancing and evolutionary epistemology. Am. Nat. 11: 187-189.

Hill, P. S. M., Wells, P. H., and Wells, H. (1997). Spontaneous flower constancy and learning in honey bees as a function of colour. Anim. Behav. 54: 615-627.

Kak, S. C. (1991). The honey bee dance language controversy. Mankind Q. 31: 357-365.

Lakatos, I. (1970). Falsification and the methodology of scientific research programmes. In Lakatos, I., and Musgrave, A. (eds.), Criticism and the Growth of Knowledge, Proceedings of the International Colloquium in the Philosophy of Science, London, Vol. 4, pp. 91-195.

Latour, B. (1987). Science in Action: How to Follow Scientists and Engineers Through Society. Open University Press, Milton Keynes, England.

Lindauer, M. (1961). Communication Among Social Bees, Harvard University Press, Cambridge, MA.

Lindegren, C. C. (1966). Cold War in Biology, Planarian Press, Ann Arbor, MI.

Medawar, P. B. ([1979] 1981). Advice to a Young Scientist, Harper and Row, New York.

Michelsen, A. (1993). The transfer of information in the dance language of honeybees: Progress and problems. J. Comp. Physiol. A 173: 135-141.

Michelsen, A., Andersen, B. B., Kirchner, W. H., and Lindauer, M. (1989). Honeybees can be recruited by a mechanical model of a dancing bee. Naturwissenschaften 76: 277-280.

Michelsen, A., Andersen, B. B., Storm, J., Kirchner, W. H., and Lindauer, M. (1992). How honeybees perceive communication dances, studied by means of a mechanical model. Behav. Ecol. Sociobiol. 30: 143-150.

Michener, C. D. (1974). The Social Behavior of the Bees: A Comparative Study, Harvard University Press, Cambridge, MA.

O'Dea, J. D. (2000). Why do honeybees dance? http://naturalscience.com/ns/articles/01-13/ns_jdo.html

Platt, J. R. (1964). Strong inference. Science 146: 347-353.

Popper, K. R. (1957). Philosophy of science: A personal report. In Mace, C. A. (ed.), British Philosophy in the Mid-Century, Macmillan, New York.

Ribbands, C. R. (1953). The Behaviour and Social Life of Honeybees, Bee Research Assoc. London.

Rosin, R. (1980). The honey-bee "dance language" hypothesis and the foundations of biology and behavior. J. Theor. Biol. 87: 457-481.

Schmidt, J. 0. (1994). Attraction of reproductive honey bee swarms to artificial nests by Nasonov pheromone. J. Chem. Ecol. 20: 1053-1056.

Schmidt, J. 0. (1999). Attractant or pheromone: The case of Nasonov secretion and honeybee swarms. J. Chem. Ecol. 25: 2051-2056.

Seeley, T. D. (1985). Honeybee Ecology: A Study of Adaptation in Social Life, Princeton University Press, Princeton, NJ.

Seeley, T. D. (1995). The Wisdom of the Hive: The Social Physiology of Honey Bee Colonies, Harvard University Press, Cambridge, MA

Steinbeck, J. ([1941] 1962). The Log from the Sea of Cortez, Viking, New York.

Towne, W. F., and Gould, J. L. (1988). The spatial precision of the honey bees' dance communication. J. Insect Behav. 1: 129-155.

Vadas, R. L., Jr. (1994). The anatomy of an ecological controversy: Honey-bee searching behavior. Oikos 69: 158-166.

Veldink, C. (1989). The honey-bee language controversy. Interdisc. Sci. Rev. 14: 166-175.

von Frisch, K. (1937). The language of bees. Sci. Progr. 32: 29-37 (reprinted in Wenner, 1993).

von Frisch, K. (1943). Versuche uber die Lenkung des Bienenfluges durch Dufstoffe [Experiments on directing bee flight by odors]. Naturwissenschaften 31: 445-460. [Translation by W. J. Nolan: Library of the International Bee Research Association.]

von Frisch, K. (1947). The dance of the bees. Bull. Anim. Behav. 5: 1-32.

von Frisch, K. (1948). Solved and unsolved problems of bee language. Bull. Anim. Behav. 9: 2-25.

von Frisch, K. (1950). Bees: Their Vision, Chemical Senses, and Language, Cornell University Press, lthaca NY.

von Frisch, K. (1962). Dialects in the language of the bees. Sci. Am. 207: 78-87.

von Frisch, K. (1967). The Dance Language and Orientation of Bees, Harvard University Press, Cambridge, MA

Waller, G. D. (1970). Attracting honeybees to alfalfa with citral, geraniol, and anise. J. Apicult. Res. 9: 9-12.

Weidemuller, A., and Seeley, T. D. (1999). Imprecision in waggle dances of the honey bee (Apis mellifera) for nearby food sources: error or adaptation? Behav. Ecol. Sociobiol. 46: 190-199.

Wells, P. H. and Wenner, A. M. (1971). The influence of food scent on behavior of foraging honeybees. Physiol. Zool. 44: 191-209.

Wells, P. H., and Wenner, A. M. (1973). Do bees have a language? Nature 241: 171-174.

Wells, P., Wells, H., Vadehrav. Vu, N., Lee, C., Han, R., Han, K., and Chang, L. (1993). Does honey bee Nasanov pheromone attract foragers? Bull. South. Calif Acad. Sci. 92: 70-77.

Wenner, A. M. (1959). The relationship of sound production during the waggle dance of the honey bee to the distance of the food source. Bull. Entomol. Soc. Am. 5: 142.

Wenner, A. M. (1962). Sound production during the waggle dance of the honey bee. Anim. Behav. 10: 79-95.

Wenner, A. M. (1964). Sound communication in honey bees. Sci. Am. 210: 116-124.

Wenner, A. M. (1971). The Bee Language Controversy: An Experience in Science, Educational Programs Improvement Corp., Boulder, CO.

Wenner, A. M. (1992). Swarm movement: A mystery explained. Am. Bee J. 132: 27-31.

Wenner, A. M. [with K. von Frisch] (1993). The language of bees. Bee World 74: 90-98.

Wenner, A. M. (1997). The role of controversy in animal behavior. In Greenberg, C, and Tobach, E. (eds.), Comparative Psychology of Invertebrates: The Field and Laboratory Study of Insect Behavior, Garland, New York, pp. 3-37.

Wenner, A. M. (1998a). Honey bee "dance language" controversy. In Greenberg, C., and Hara, M. (eds.), Comparative Psychology: A Handbook, Garland, New York, pp. 823-836.

Wenner, A. M. (1998b). Odors, wind and colony foraging - Part I. The need for odor. Am. Bee J. 138: 746-748.

Wenner, A. M. (1998c). Odors, wind and colony foraging - Part II. The role of wind direction. Am. Bee J. 138: 807-810.

Wenner, A. M. (1998d). Odors, wind and colony foraging - Part III. Insights from beehunting. Am. Bee J. 138: 897-899.

Wenner, A. M., and Wells, P. H. (1990). Anatomy of a Controversy: The Question of a "Language" Among Bees, Columbia University Press, New York.

Wenner, A. M., Meade, D., and Friesen, L. J. (1991). Recruitment, search behavior, and flight ranges of honey bees. Am. Zool. 31: 768-782.

Wilson, E. O. (1971). The Insect Societies, Harvard University Press, Cambridge, MA.

Winston, M. L. (1987). The Biology of the Honey Bee, Harvard University Press, Cambridge, MA.

Winston, M. L., and Slessor, K. N. (1993). Applications of queen honey bee mandibular pheromone for beekeeping and crop pollination. Bee World 74: 111-128.