On an Unfortunate Seasonal Confound in the Study of Gremlin Cohabitation

A later attempt to reproduce Merrow’s cohabitation study produced results so dramatically at odds with the original that, for approximately forty-eight hours, I entertained the possibility that our first conclusions had been wrong. They were not. We had simply forgotten what month it was.

The repetition was undertaken six years after the original experiment by a younger research group at the Collegium of Bellcross, whose principal aim was to determine whether the earlier findings would hold among Gremlins drawn from a different urban population. I was invited as an adviser, partly because of my involvement in the first study and partly, I suspect, because the Collegium wished to attach an older name to a younger grant.

The design was sensible and the accommodations improved, with consent procedures clearer than ours had been. Participants were screened for prior acquaintance, occupational conflict, health concerns, and several other variables we had neglected the first time. The researchers were, in short, considerably more organised than we had been, which made what followed especially satisfying to me once it ceased being alarming.

The first group consisted of six adult Gremlins. Under our previous observations, six was a difficult number, and one would expect partitioning, territorial rearrangement, improvised barriers, staggered schedules, and at least one participant deciding that a broom cupboard constituted an acceptable private residence. None of this occurred.

Within the first day, the six were spending nearly all their time together, eating and working in company and talking incessantly. One participant who had been assigned a private sleeping room dragged his bedding into the common hall because, as he explained, “everyone else is here.” This was so profoundly unlike the behaviour documented in our first study that the research team became ecstatic.

Doctor Pellin, who led the repetition, immediately suggested that urban Gremlin sociality might vary by region. One assistant proposed that improved accommodations had removed the stressors responsible for avoidance, and another wondered whether the original participants had been unusually antisocial. I advised patience, which proved wise.

On the second day, cooperation increased further, but so did attention of a different sort. Grooming behaviour became common, and participants began exchanging small objects, food, bits of jewellery, tools, buttons, and things which did not appear to belong to any of them. Two males spent nearly an hour polishing one another’s belt buckles. A female dismantled a mechanical pencil and presented the spring to another participant with great ceremony, and he accepted it as though she had handed him a medal.

At this point I asked whether anyone had checked the seasonal calendar. There was a silence. Doctor Pellin looked at me, I looked at Doctor Pellin, and one of the assistants said, “Oh.” The Gremlin mating season was due to begin within the week.

In our defence, Gremlin reproductive timing is not perfectly regular. Onset varies with climate, nutrition, population density, and several physiological factors which remain poorly understood, and in Bellcross that year it had apparently begun early.

The implications were immediate. The original study had measured Gremlin cohabitation under ordinary adult conditions, in which prolonged proximity is generally aversive, whereas we were now observing the brief annual period in which that aversion diminishes sharply and, in some individuals, reverses. The six participants had not become unusually compatible. They had become extremely interested in one another.

Doctor Pellin wanted to terminate the trial. The participants objected, which produced a methodological debate of some complexity, because all six were competent adults, all had consented to the study, and all insisted they wished to remain. One asked whether the researchers planned to terminate every experiment the moment the subjects began enjoying themselves. This was unfair, and it was also difficult to answer quickly.

We therefore modified the protocol instead. Additional privacy screens were installed and medical staff placed on call, and participants were reminded repeatedly that they could withdraw at any time and that participation did not constitute consent to anything beyond the study itself. They appeared to find this explanation obvious. We did not.

The social structure changed by the hour. Ordinary Gremlin cohabitation tends toward separation, while this group tended toward convergence, competition, display, alliance, jealousy, reconciliation, and then convergence again. Possessions became socially charged: in the first study, a borrowed tool might produce an argument over ownership, and in this one the same act could apparently constitute flirtation. Unfortunately, we did not know which tools.

This became clear when one researcher attempted to return a small wrench to its owner after finding it in the common room. The owner became deeply embarrassed, and another participant laughed so hard he fell from a chair. We recorded “context-dependent exchange behaviour” and moved on.

The most remarkable change appeared when the study expanded to twelve participants. Under normal conditions, twelve Gremlins barely functioned as a household, dispersing, avoiding common spaces, and spending increasing amounts of time outside the assigned residence. This time all twelve remained, and for three days they produced something resembling a social club designed by people with no prior experience of clubs.

Furniture migrated constantly and meals became communal, while arguments erupted and vanished with extraordinary speed. Two participants who appeared to dislike one another intensely spent an entire afternoon repairing the same clock. Another pair constructed a partition, then sat on the same side of it. Several informal sleeping arrangements developed and dissolved without any pattern the researchers could reliably map.

Doctor Pellin attempted to create a chart, which became unreadable. He then created a second chart showing only preferred proximity, and that became worse. I recommended abandoning the charts; he did not, and by the fourth day the wall of the observation room displayed enough intersecting lines to resemble a railway map designed by a lunatic.

What fascinated me most was that aggression did not disappear. It changed character. The same Gremlins who, outside mating season, might have avoided one another to prevent conflict now sought one another out and tolerated remarkably intense disagreement without permanently separating. There was shouting, there was posturing, and there were several minor fights, yet the participants repeatedly returned to the same social spaces. Ordinary avoidance had become temporary conflict followed by renewed association. One male was struck in the face by another, stormed out, returned forty minutes later with pastries, and distributed them to the entire group. No one considered this unusual. I did.

At twenty participants, the experiment ceased to resemble the earlier study in almost every meaningful respect. The warehouse did not subdivide into individual territories; it became crowded, and then very crowded. Gremlins who would ordinarily have regarded continuous proximity as intolerable packed themselves into workshops, kitchens, corridors, and common rooms. Social clusters formed and dissolved continuously, and participants moved between them with such frequency that no stable division emerged.

One researcher described the pattern as “fluid social aggregation.” A technician called it “absolute bedlam.” Both were correct. The noise alone was extraordinary, since Gremlins are not generally loud people when living among other races, yet in large mating-season groups they apparently become incapable of having one conversation where five will suffice.

At one point I counted seven simultaneous arguments occurring in the same room, of which I could identify the subject of three. One concerned food, another a broken music box, and the third whether a particular participant had deliberately sat beside another participant or merely happened to sit there. That third argument lasted nearly an hour.

By the end of the week, we had discovered something more interesting than the replication we originally intended. Gremlin aversion to one another is not a fixed social incapacity but a regulated behavioural state. For most of the year proximity is costly and Gremlins minimise it, while during mating season whatever mechanism ordinarily drives dispersal weakens sufficiently to permit exactly the opposite pattern: dense social contact, competition, courtship, temporary bonding, and repeated reassociation. This is biologically sensible, since a species cannot simultaneously maintain strong adult dispersal and reproduce unless one of those behaviours periodically yields to the other. Obvious in retrospect, as most useful observations are.

The experiment was terminated two days earlier than planned. Officially, this was because hormonal and behavioural changes had rendered comparison with the original study impossible. Unofficially, it was because Doctor Pellin entered the common room on the ninth morning, observed the situation for approximately ten seconds, returned to the observation station, and said, “I no longer believe we are studying cohabitation.” I asked what he believed we were studying. He said, “Consequences.” I considered this imprecise, and also fair.

The most embarrassing detail is that the seasonal error had been preventable. The date was present in the project calendar from the beginning, and one of the assistants had even marked the estimated start of mating season in red ink. No one noticed. Years later, Doctor Pellin continued to refer to the event as “an unforeseen physiological confound.” It was not unforeseen. It was written on the wall.