mind swarm 1

The Mind of the Swarm
David Pacchioli

Research/Penn State, Vol. 19, no. 3, September, 1998
(With the kind permission of Nancy Marie Brown, Director, Research Publications, Penn State University)

Cactus City wasn't hard to find. I just followed the 18-wheelers east from Indio until the dun-colored desert had snuffed every rumor of sprinklers and turfgrass, and kept on going. Past the last billboard for a tribe-owned casino. Past the relay of concrete cisterns stenciled "Water for Radiator Use Only." Past a bleached rest stop, scorpion's paradise, to a Twilight Zone of a turnoff; down a short stretch of cracked blacktop to a lonely quadrangle of chain-link fence protecting a thicket of pipework for the Southern California Gas Company. From that point I was on my own.

Mindful of my rental car, I crept half a mile of unpaved road at walker's pace, steering around the largest rocks and the softest-looking patches of sand, looking for signs of life. The harsh afternoon light was mellowing rapidly and I was beginning to wonder whether I was going to have a Cactus-City sundown all to myself when a two-legged figure stepped from the scraggle of creosote bushes ahead, a walkie-talkie held up to its face. It waved me forward, disappeared for a moment back into the brush, then re-emerged as I approached: a slight man in purple t-shirt, baggy pants, and Birkenstock sandals. His prominent ears, close-cropped hair and goatee, and dark glasses completed a hipster effect. He stuck his hand through the open driver's-side window, beaming. "You made it," Scott Camazine said. "Welcome to Cactus City. Leave it here and follow me."

I got out of the car and followed, stiff from traveling, through a maze of stout cholla cactus and tall, fingerlike ocotillo toward the rounded crown of a paloverde tree. In front of the tree, when we reached it, was a beach chair and a lightweight table; behind these a large man stood squinting at the back of what looked like a rustic telephone stand: A rectangle of sturdy plywood mounted on a wooden post and hooded with Plexiglas. The man was Kirk Visscher, associate professor of entomology at the University of California at Riverside, and Camazine's research partner and friend. The dark mass at the center of the plywood was not a telephone but a densely packed, football-shaped swarm of 5,000 honeybees. "Here we are," Camazine said.

mind swarm 3

The swarm was utterly quiet, tanked up on sugar-water from the tiny trough that Visscher was now refilling with a plastic squeeze bottle, and settled in for the night around its queen, who was secured in a small cage hidden under layers of her workers. But the accommodations were only temporary: the stand was too exposed to be more than a way station. A good place, it was, for watching a swarm of bees in the process of searching for a new home.

From a scientist's perspective, the day had not gone well. Visscher and Camazine, assistant professor of entomology at Penn State, had been out with their clipboards and their walkie talkies since early morning, and had seen very little in the way of apian activity. The bees were taking their own sweet time about moving, and the humans looked dusty and tired. "This is one of the most desolate places I've ever been," Camazine groused mildly. "Nothing lives here." He grinned. In fact, it was mid-March, coming off a rainy El Niño winter, and the desert was in rare, albeit subtle, bloom. The ocotillo were tinged with crimson, and tiny purple flowers sat like gems on the sand. Delicate California poppies added dashes of bright yellow. Lizards scurried underfoot, and the light breeze carried bird calls. "You won't see it again like this for a hundred years," Visscher said. There was evidence of human presence too: shell casings and fragments of things shot at; shards of plastic, blasted beer cans. In the distance, far enough to be scarcely audible, was the steady stream of trucks on Interstate 10, bound for Phoenix or L.A. Beyond, dwarfing everything, the rugged ocher hulk of the Orocopia mountains.

What there wasn't, out here in Cactus City, was an abundance of potential homesites, especially for Apis mellifera. There were no large trees with cozy hollows in this flat and open landscape, no significant cavities in the ground; nowhere, in short, for a swarm of honeybees to hole up and feel secure. That was precisely why Visscher and Camazine had chosen this spot. It was perfect for testing their hypothesis about honeybee behavior. The only suitable nest sites for miles around were the ones the researchers had set out themselves, two wooden hive boxes they had brought from Riverside, 90 miles west. They had placed one box 100 yards to the north and the other an equal distance to the south of where we were now standing. With an observer at each box and one at the swarm, they could monitor the househunting process as it unfolded.

The choice of a new nest site by a swarm of bees is one of the clearest examples of group decision-making outside of human behavior," Camazine and Visscher have written. In the wild, the process begins when a colony divides: The group having grown too large to sustain itself, the queen and several thousand of her workers break away. The runaway swarm settles, temporarily, on a tree branch or similar perch, and from there scouts fly out to reconnoiter. When a scout locates a suitable site for a nest, she (all scouts, all workers, are female) returns to the swarm and reports her find, communicating in the dance language whose code was broken by Austrian Nobel laureate Karl von Frisch in the 1940s. Other scouts, meanwhile, return to dance for other sites: the better the site, the more vigorous the dance. Rival sites gain adherents, as dance followers are recruited and in turn dance for others. There are tidal surges of popularity — for one place, then another, then another, as reports come in. At some point, however, in a matter of hours $#151; or days — things settle out, and a consensus is reached. The dancing gradually becomes unanimous. All the bees return to the swarm. There's a brief period of calm, and then activity of a different sort builds to a crescendo: scouts strafe the swarm in a behavior known as buzz-running. Finally, obeying some hidden signal, the swarm takes off and flies en masse to the chosen site. Human observers talk excitedly of running to keep up with a moving cloud of bees.

"The most distinctive feature of this selection," Visscher and Camazine write, "is that it is a collective decision. Few if any of the bees have an opportunity to acquire information from multiple sites on the many variables affecting the choice. Yet, as a group, the bees of the swarm come to an agreement." Not only that: They consistently manage to decide wisely, choosing a site that is well-suited for their survival. The question is how. How do bees find and evaluate sites? How do they choose among alternatives? How do they come to agreement, and how do they know when an agreement has been reached?   

mind swarm 2

The answers to these questions, to a large extent, remain locked in the mind of the swarm. There's only one way to approach them, really, and that is to watch bees in action until you think you understand. Cleverly plotted experiments expedite things, but however you slice it, being a bee behavioralist means large doses of watching and waiting, and lots of painstaking labor. Last year, for one experiment, Camazine and Visscher glued tiny numbered tags to the thoraxes of 400 individual bees, returned these subjects to the swarm, videotaped their activity for a total of 12 hours, and then, with the help of a loyal team of assistants, ran and re-ran that video through enough single-frame playbacks to discern and transcribe every last bee's predominant behavior at one-minute intervals.

Not that there aren't surprises to keep things interesting. "That's the thing about bees," Camazine said over a burrito dinner in Harvey, the creaky 25-year-old mobile home where he and Visscher roost on desert trips. "No matter how much you think you know them, they always do something you don't expect." To put it more succinctly, quoting Winnie the Pooh, "You never can tell about bees." The honey-loving bear's lament is the bee scientists' motto. Today, for instance, the swarm simply hadn't budged, no telling why. And last year, Camazine remembered, after tagging those 400 bees, "I went out to the stand one cool morning and not one of them was visible on the swarm. Turned out they had all burrowed underneath. The rest of the bees were keeping them warm."

After dinner, we stepped out behind Harvey to where Visscher had set up a good-sized amateur telescope, a 10-inch refractor, for the evening's entertainment. The moon was so bright I could read my watch; that and a scattering of clouds made it impossible to see much in the way of stars. Before turning in I followed Camazine out to the swarm stand, where he fed the bees their bed-time sugar-water. "It's important to keep them well fed," he said, "so they aren't tempted to forage for food. We want their only concern to be finding a home."

Dawn broke orange across the desert at 5:45. By six, Camazine and I were tromping through the sand to check on the bees. The morning was calm, but the night had been windy enough to rattle Harvey's windows, and to make me feel sorry for Visscher, who had chosen to sleep outside. Montana-born, he wouldn't be cold, Camazine had assured me. ("We have an ongoing battle about temperature, believe me.") But Visscher did concede that the stubborn gusts had wakened him several times. Once or twice he had gotten up, and been afraid his tent would blow away while he was out of it — which would have been fitting, he admitted, since that was how he had got the thing in the first place: "It blew into the bee yard one day, sleeping bag and all. I put an ad in the paper, but nobody claimed it."

I had to hustle to keep up as Camazine made the rounds. The nest boxes, north and south, had been blown from their perches. The swarm stand was turned at a 45-degree angle from where it had faced yesterday. But the bees themselves were intact, huddled tight in the middle of their plywood roost. "Look," Camazine said. "Some are already starting DVAV." Sure enough, about a dozen bees on the surface were shaking their tiny abdomens like they really meant it, a behavior known as dorsal-ventral abdominal vibration. "It's kind of a wake-up call," Camazine explained: "Get with the program." It was an encouraging sign, but on the way back to Harvey Camazine still seemed worried. "It's all up to the bees now," he said; and after a sudden gust he added, "They won't fly if it's this windy."

The classic studies of apian house-hunting were initiated by the German Martin Lindauer during the early 1950s. Reporting on his observations of 19 swarms loosed on the outskirts of Munich, Lindauer drew several important conclusions about bees" aptitude for decision-making. The most startling was this: Individual bees actually compare sites. Presented with a strong-enough argument for another site, they can be persuaded to drop their favorite. They change their minds.

For a creature with a nervous system that packs a mere one million neurons — hardly enough firepower to warrant the term "brain" — evaluating one home site is impressive enough. There is convincing evidence that scouts are able to assess several characteristics in a potential dwelling place: its size (they prefer a volume of about 40 liters), its freedom from drafts, and the size of its entrance (smaller is easier to defend). Comparison, however, is a quantum leap above this. To choose between sites, a scout would have to create a mental image of a potential home, hold more than one such image in its tiny memory, choose the best among these, and communicate that choice to the swarm.

In recent years, some animal behavioralists have seized on Lindauer's hypothesis as a vivid example of high-level cognitive ability in social insects. Some have even compared the bee's house— hunting technique to the process by which humans buy real estate: extensive consideration and lots of shopping around. "The trend today is to think that we've underestimated animal awareness," Camazine had told me during dinner. He and Visscher are not so sure.

It's tough enough to know what a fellow human is really thinking; how much more inscrutable, then, are the recesses of an insect's mind? Even as graduate students — at Cornell during the early 1980s, working in the lab of Thomas Seeley, a renowned expert on honeybee behavior —Camazine and Visscher found Lindauer's evidence too thin. Lindauer had drawn on the detailed histories of only three bees to make his case. His work begged for follow-up — and perhaps the question needed rephrasing. Instead of proving how much awareness a honeybee possesses, wouldn't it be easier to see how little it needs to accomplish a task? What's the bottom line? Or, as Camazine puts it, "How dumb can you be and still get the job done? This is the question we're trying to answer."

Camazine and Visscher had arrived at Cornell by different routes. Both were undergraduate biology majors at Harvard during the late '70s. Visscher's subsequent path into bee science, however, seems to have been straightforward. Both his grandfather and his father — by profession a marine biologist and a physician, respectively — had been hobby beekeepers. So had Visscher himself, growing up in Montana.

Camazine, on the other hand, while admittedly bug-obsessed from a tender age, took a lengthy detour via Harvard Medical School. "I always wanted to do research," he explains, "maybe study medical entomology, or parasitology, with a natural history approach. But I figured with a medical degree I could do anything a Ph.D. in biology could do, and not vice versa. I figured there would be no down-side. I was also interested in medicine, of course." From the first, he sought out research opportunities, working first in an immunology lab, then in neurobiology. After getting his M.D., he completed a surgical internship. "Then I just peeled off," he shrugs. He did work in ethnopharmacology, studying the medical practices of Zuni Indian healers in New Mexico, and in chemical ecology, identifying various kinds of plant-and-animal interactions. Finally, when he decided to go for a Ph.D., Camazine supported himself by working part-time as an emergency-room physician. "It gave me the freedom to do what I wanted in grad school," he says, of his medical training. "Now it mostly comes in handy for things like looking in my daughter's ears." That and helping out with emergencies on airplanes. As a recent flight landed in California, Camazine admits, he attended a woman who had fallen and dislocated her knee. ("I popped it back, and then I got the heck out of there.")

For his thesis at Cornell, Camazine had investigated a basic biological mystery: the pattern in a honeycomb. The inside of a beehive is one of nature's wonders, made up of a series of parallel wax combs, each containing thousands of hexagonal cells which worker bees busily fill, either with food or with offspring. To the human eye, the combs exhibit a uniform pattern. In cross-section, the brood nest — eggs, larvae, and pupae — is located at the center of the comb, surrounded by a rim of pollen and then a larger circle of honey around the perimeter. The regularity of the pattern, Camazine writes, suggests its importance to the colony's survival, perhaps related to incubation temperature or feeding efficiency. It would seem to be an evolutionary adaptation. But what mechanism, what intelligence, brings it about? The standard explanation — that there must be a sort of blueprint of the comb in the mind of every bee, that bees are programmed to know what a filled comb should look like, and what goes where — didn't cut it with Camazine. For one thing, he thought, it couldn't explain the aberrations that occur from time to time, the honey or pollen occasionally misplaced in the midst of the brood nest. Finally, after coming across the work of a Belgian chemist named Jean-Louis Deneubourg, he came up with an alternative explanation.

Deneubourg, a student of Belgian Nobel laureate Ilya Prigogine, was then in the forefront of a new field of study, that of so-called self-organized systems. Beginning in chemistry, self-organization was a concept that sought to explain some of the mysterious order evident in nature— the ripples in a sand dune, say, or a zebra's stripes — in terms of simple interactions between the components of the system in question. In the case of a sand dune, for instance, the characteristic pattern is not imposed by some external ordering force; it emerges spontaneously from the play of particles of sand affected by wind and gravity.

Certain biological systems, Deneubourg reasoned, while more complex than chemical or physical systems, might follow the same sort of process. The elegant order in a school of fish, for example, might be achieved not by the expert guidance of a leader, nor by each fish possessing the "global" idea of how a school must look, but simply by the combined interactions of individual fish obeying a very basic, "local" rule: I must gauge my direction and my velocity so as not to bump into, or stray too far from, my neighbors.

For his thesis, Camazine created a mathematical model applying self-organization to the concentric pattern of the honey-comb. The crux of the model was a set of simple behavioral rules gleaned from observing actual hives: One of these was that the queen tends to lay her eggs in cells that are close to one another; a second was that the foodstuffs stored closest to these brood cells are removed for eating more readily than those stored farther away. In a system governed by simple rules such as these, he showed, even though queen and workers deposit eggs, honey, and pollen without a thought for global order, eventually the concentric pattern will emerge.

By the time Camazine was finishing at Cornell, Visscher had moved on to Riverside. Both researchers were involved with numerous other bee-related projects, but the househunting question continued to intrigue them. Visscher's location near the desert would make experiments easier, and Camazine's self-organization work suggested that modeling would be a valuable complement. In 1992, Camazine did some preliminary swarm observations in upstate New York, in a forested area of the Adirondack mountains known as Cranberry Lake. Last year, they began work in Cactus City.

In that first set of desert experiments, Camazine remembers, he and Visscher documented the househunting activity of 80 scout bees. Of these, he says, "only three or four changed their minds," dancing first for one site and then for another, as Lindauer's scouts had done. "That really made us question the importance of comparison to the decision-making process." This year, the two devised an experiment to prevent comparison from taking place: They removed any scouts who "crossed over," from one site to another, and noticed whether their absence had any effect. After three experimental trials and three control studies, they concluded that decisionmaking was not affected — not even delayed. "Everything was identical," Camazine says. "But that raises another question. If comparison is not important to the swarm's decision, why are some bees doing it? For that matter, why do some scouts keep going back to a nest site after they've already checked it out?"

After breakfast we picked up our walkie talkies and our plastic caddies, each of the latter including an electronic data collector (a sort of glorified calculator), a fine-mesh bee net, and two tiny bottles of paint, with brushes. Then we fanned out to our positions: Visscher at the north box, Camazine at the swarm stand, and me at the south box, where I would sit in briefly for Rick Vetter, Visscher's laboratory technician, who was on his way from Riverside.

The three had brought the swarm out the day before yesterday, and placed it on the stand late that afternoon to give the bees a chance to get acclimated. For these purposes, a swarm is a pound and a half of bees, including a queen, pulled from one of the colonies hived at the beeyard, placed in a small screened box, and fattened on sugar water for five days. Being engorged with this "nectar" stimulates the bees" wax glands, making them eager to find a place where they can build comb and store honey.

All day yesterday Visscher and Camazine had waited. So far, however, this swarm's scouts had completely failed to locate the south box. Two bees had managed to find the north box, and had been duly netted, daubed with pink paint, and released. "Tomorrow morning we hope they'll go back to the north box, recruit others, and get the whole process going," Camazine had said at dinner. According to form, he explained, there should be some real action by ten or 11:00. By noon, there would be full-blown dancing on the swarm, and visits to the nest boxes would become so numerous they would be hard for the observers to keep up with. By three or four, the show might be over, the swarm busily settling into its new digs. But things didn't always go according to form. Scouts from the last swarm they had released, Camazine admitted, instead of heading off to one of the boxes, had found their way to one of the mobile home's external drain spouts, which he and Visscher had forgotten to close up. Lindauer himself had reported, back in the '50s, that two of his 19 swarms had taken off without coming to consensus, broken up in mid-flight, and eventually straggled back to the stand. One swarm had not taken off at all.

The south box was dead quiet. The sun was still low, and sitting in my beach chair I could feel the coolness of the sand. The morning's silence was marred only by the occasional buzzing of a fly. Standing, I could see the silent stream of trucks up on the interstate, but in the foreground Camazine and Visscher were hidden from view. Soon, Camazine's voice broke onto the walkie talkie: "There's lots of activity here at the swarm stand," he said. "Lots of DVAV, lots of flying around." His voice sounded hopeful. Then, more silence.

In the early trials at Cranberry Lake, I had read, scouts had located over 40 suitable nest sites in the surrounding forest, some as far as three miles away. How did they cover so much territory, I wondered, and how thoroughly? "We'd love to be able to follow a scout as she makes her rounds," Camazine had said, "but nobody's figured out a way to do that yet." He said it as though a bee-borne minicam was only a matter of time. And why not? Back at Penn State he had once devised a sort of tread-mill for measuring a bee's energy output, a tiny "merry-go-round" apparatus to which he could attach a single bee. Had scrapped it, though: too tough to use.

At 8:00 a plume of dust appeared in the direction of the access road, and soon the white top of a Chevy Suburban became visible, headed toward Harvey. Five minutes later Rick Vetter appeared in person, a bounding, athletic type sporting a floppy desert hat and a graying fu manchu, and carrying a black banjo case. We said hello and then, relieved of my post, I followed the trail of blue ribbons that had been tied in bushes leading to the swarm. When I got there, Camazine was slumped in his chair at the foot of the stand, a t-shirt draped over his head for sun protection; he looked vaguely Pharaonic. "Here you go," he said, rising. "I'll show you some dancing."

The dancing bees, half a dozen, were rotating in place, counter-clockwise on the surface of the swarm, pausing at the same spot in each circuit to execute an exaggerated abdominal waggle. A score of other bees followed each one intently, pushing forward, nudging their heads close to the dancer's rear, "reading" the dance by sticking their antennae into the space directly behind the dancer's wings. "That's where they can best perceive the sound of the wings' vibration," Camazine said. "They need to follow several circuits in order to get the message." Dancers signal the direction of a site by the way they face when they waggle, he went on, orienting themselves by the sun if they can see it, or by gravity if they are in shade. They indicate distance by the alacrity of their circuits: fast means a site is nearby; slow means it's farther away. The bees dancing now were facing about 5:00 from the position of the sun — roughly the direction of the north box, by Camazine's calculation — but their circuits were relatively slow, each requiring about three seconds to complete. "They're probably dancing for a site in those mountains over there," Camazine said, pointing. Then he culled these "rogues," plucking them off the swarm with thumb and forefinger and dunking them through the plastic lid of a Del Taco soft-drink cup. "We don't want them to interfere with what we're trying to look at," he said.

I suggested that the interpretation of direction, especially, seemed a tricky business — full of mental gymnastics and compound geometry. "In my animal behavior class," Camazine acknowledged, grinning, "I have the students get in a circle and take turns, each one does a waggle dance. The others have to tell what location is being danced for. It's not that easy for them to get."

He sat back down. "It's a little anxiety-provoking, wondering if they'll find the boxes," he said, almost to himself, "but there's nothing we can do now except wait."

Over at the north box, Visscher was working on the computer program that he and Camazine were writing: their simulation of the house-hunting process. Every once in a while, his voice broke in on the walkie-talkie with a question. Talking about the simulation perked Camazine up again. "What we think is going on with decision-making," he said, "is a self-organized process. Scouts will check out a site, dance for a few minutes, then stop. During that time other bees are recruited, and maybe they will dance. The dancing goes on longer for the better sites, though, and pretty soon you get a positive feedback going. Then it just snowballs. The best sites are continually being advertised, gaining recruits, while the not-so-good ones are losing ground.

"In this system, no bee has to compare multiple sites. Each one goes to a certain site, behaves a certain way — she dances or she doesn't — and by this system of positive feedback a decision emerges. The relative merit of alternatives is determined blindly. "If it is like that, a matter of individuals following simple rules," Camazine said, "then it's an ideal process to model on a computer. We set up the rules, plug in values from our experimental data, and let it run. If it works — if simple rules are sufficient — it won't prove that this is how bees decide, but it will show us that they could do it this way."

At 9:04, according to the small clock fixed to the front of the swarm stand, Visscher radioed: "Bee in the box." "Hallelujah!" Camazine countered. At 9:06, Visscher reported that he had netted and marked the visitor. At 9:11, paint barely dry, she arrived back at the swarm. Camazine peered intently to see if she would dance, but instead she wandered slowly across the surface of the swarm, paused a second, then burrowed herself into its middle, disappearing from view. Camazine heaved an exaggerated sigh. A few seconds later the pink scout re-emerged, and resumed wandering, but she seemed in no mood to dance. "There's a lot of chance involved," Camazine said. "You figure, fewer than one in ten bees is a scout — about 400 of 5,000. Of those, maybe one in ten finds the box. Then, maybe one in ten of those comes back and dances. It's hard to get the whole process started. Once you do, though, it really takes off."

Fifteen minutes later, Vetter reported that a bee had checked in at the south box, where I had begun the morning. Soon he had another. Things seemed to be looking up. There was banter on the walkie-talkies. When the high-pitched burr of a cell phone interrupted, Camazine slipped the phone from his pocket and unfolded it. "Swarm Central," he said brightly. "How may I direct your call?" The flurry was short-lived, however. An hour later, things had died out. Neither of Vetter's scouts had showed up back at the swarm. We were bee-calmed; and the sun was starting to get hot.

The trail to Visscher and the north box was easy to follow: I just walked toward the thrum of a Honda generator, then followed its cord to Visscher's computer, which sat in his lap in the small patch of shade created by a single small paloverde tree. The white nest box sat in the crotch of the tree where Camazine had re-wedged it that morning, weighted this time by a length of stout branch and a large rock.

Like Camazine's, Visscher's work with bees ranges widely. Two years ago, he attracted international media attention with an experiment in which he and Vetter submitted themselves to numerous stings — Visscher estimated he took 70 in all — in order to test the prevailing wisdom about the best method of sting removal. ("What we found is that it doesn't make a difference whether you scrape or pinch," Visscher reported. "What makes a difference is when you do it.") Other experiments have addressed issues of conflict of interest in honeybee colonies, and the evolutionary significance of cooperation. Visscher's feeling for bees is obviously profound. At one point he said quietly, choosing his words, "There is always this contrast between complexity and flexibility on the one hand — sometimes it seems like bees can do anything — and evidence of very simple kinds of interactions." Just then Camazine's voice broke in on the walkie-talkie. "I have three bees dancing vigorously at 5:00," it said. "Could that be for your box, Kirk?"

"Time complete circuits," Visscher responded. Then he stood, set down his computer, and headed for the swarm.

"It's 15 to 20 seconds for five circuits," Camazine said, when we arrived. That was too long. "The ones we want would be ten seconds," Visscher confirmed. There followed some thinking out loud. "Maybe our boxes don't smell good," Camazine said. They were converted supers, or hive boxes; bees had once lived in them, and had no doubt left traces of their wax. But these particular boxes had been empty for at least a year. Blaming the boxes, Visscher acknowledged, was "voodoo, but it's rational voodoo. Maybe we should get fresh boxes." For the moment, though, they decided to persist with what they had. Camazine stayed at the swarm, while Visscher set off walking toward the highway. He wanted to check the gas company's fenced enclosure, three-quarters of a mile off, to see if he could spot any wayward rogues.

Meanwhile, over at the south box, Vetter was undergoing some soul-searching of his own. Maybe the lack of movement was his fault. He had made up this swarm — and done it on an unusually cold, rainy day, he said. Maybe the foul weather had had some negative effect on the character of this batch of bees. "Maybe we didn't get enough potential scouts." He stood and stretched, looking back in the direction of the stand. "You never can tell about bees," he said.

At 4:00 Camazine and Visscher held another consultation, this time meeting at the north box. Weary now, I remained at the swarm stand and waited. But for the dozen or so rogues buzzing angrily in their soft-drink cup, the bees were entirely docile, awaiting their fate. After 15 minutes I heard and then saw the two researchers approaching through the scrub, their frowns announcing their decision. "We're giving up on this swarm," Camazine said, apology in his voice. "We'll take it down tonight, and head back to Riverside in the morning. We'll come out again next week with a new one."

Toward sundown, the day's disappointment soothed a little by fresh-baked biscuits and the jaunty notes of Vetter's banjo, Visscher and I set out to hive the swarm. At the site, Visscher hoisted a wooden super onto the table. Turning to the stand, he took an envelope-sized piece of thick cardboard and, with sure strokes, swept the swarm away from the queen cage at the center of the board. A score of bees, aroused, flew up and buzzed his ears angrily, spiralling out around his head. Ignoring them, Visscher unfastened the tiny cage and held it up for me to see. Then he pulled one of the frames from the super, fastened the cage to it, and redeposited it carefully, so as not to crush the queen. Next, with the cardboard scoop, he ferried a hundred bees from the stand and shoveled them into the super. Then he stepped behind the stand, lifted it off its wooden post, carried it to the table, and banged it down sharply three times, dropping the rest of the swarm onto the aluminum surface in dense clattering clumps.

In seconds the hardiest bees had regained their bearings and picked up on an invisible signal coming from their swarm-mates already in the super. These few im-mediately set about relaying the signal to the rest, thrusting their posteriors high in the air. "They're releasing a pheromone," Visscher said. "It means, 'Come here.' It smells like lemon, if you get enough of it." I bent to smell, but couldn't detect anything.

"Watch this, now," Visscher said. "If we're lucky, we're about to see an especially beautiful behavior."

Sure enough, as the signal was passed, a wave of order spread gradually through the jumbled mass. Slowly, row by row, the bees remembered themselves, drew them-selves together, assumed their places, passed it on. Like a chemical reaction, a photograph developing, there emerged a unifying pattern: neat columns, a phalanx, an army, awakened and marching steadily, relentlessly forward. Slowly, silently, step-by- step, oblivious, the members of the swarm approached the box and spilled across its threshhold; no straying, no shoving, no stepping out of line.

Scott Camazine, M.D., Ph.D., is assistant professor of entomology in the Penn State College of Agricultural Sciences, 539 Agricultural Science and Industries Bldg., University Park, PA 16802; 814-863-1854

P. Kirk Visscher, Ph.D., is associate professor of entomology at the University of California at Riverside.

Some of the work reported above was funded in part by grants from the National Science Foundation.

Research/Penn State is published by the Vice President for Research.

Contents copyright 1998 The Pennsylvania State University, University Park, PA 16802-3303. Contact the editor for permission to reprint.