Coffee with Claude

My father subscribed to Popular Electronics. I read it religiously.

This was the early 1970s, and the magazine landed in our mailbox like a monthly dispatch from a future that was always just one soldering iron away. Every issue carried build-it-yourself plans for devices that straddled the line between useful and magical — shortwave converters, music synthesizers, substitution boxes for the hobbyist bench. The ads promised you could discover the ease and excitement of training at home. The articles promised you could build the future in your garage.

In February of 1971, the cover announced "ELECTRONICS STIMULATES Household Plant Growth," and inside, L. George Lawrence laid out an experimental arrangement for electroculture — using high-voltage electrostatic fields to enhance plant development. It was fascinating enough, but the real hook came four months later. In the June issue, Lawrence published a companion piece: "More Experiments in Electroculture," subtitled "DO THEY REALLY KNOW IF YOU CARE? FIND OUT ELECTRONICALLY." That article included complete plans for what he called a Plant Response Detector — a psychogalvanometer you could build yourself.

The device was elegant for a magazine project. At its heart sat a Wheatstone bridge, with the plant leaf forming one arm of the circuit between two electrode binding posts. A µA741C operational amplifier with a large-signal gain of 100,000 amplified the offset signal from the bridge. An audio oscillator driven by a pair of transistors converted the plant's electrical state into an audible tone whose pitch shifted with changes in leaf conductance. The whole thing had visual output on a milliamp meter, audio output through a small speaker, and binding posts for connecting a tape recorder or pen-type chart recorder to make permanent records. Lawrence even specified the electrode jelly — the same conductive cream physicians used for electrocardiograms.

I was in high school, and I was hooked. But to understand why, you need to know something about the laboratory I grew up in — which is to say, my bedroom.

My earliest memories are inextricably merged between images of catching garter snakes that were longer than I was tall and discovering that my father's soldering iron was hot enough to deliver a painful burn on the finger of a curious child. To a well-equipped child scientist, the tools of your laboratory include pencils, notebooks, beakers, test tubes, bones, butterfly net, killing jar, microscope, telescope, live animals, decaying matter, high voltage transformers, comic books, Jules Verne novels, dissected toads, tape recorders, lots of wire, and anything else that is really cool. I built my first computer in the seventh grade using more than a hundred relay switches and miniature light bulbs. By the time the electroculture articles arrived in our mailbox, I had already built an apparatus to measure the brightness of stars, a battery-powered mosquito repeller, and would soon simulate the physical environment of Mars using an old freezer, a vacuum pump, and tailings from an iron ore mine. When your father happens to be an electrical engineer, a teenager doesn't think twice about any of this. But I think about it now. My brother played little league. My father could have steered me there too — toward something ordinary and legible. Instead, he saw the scientist growing in his own house and did what any good naturalist does: he didn't interfere with what was emerging. He just made sure it had what it needed.

So the psychogalvanometer was not an isolated project. It was one more instrument in a childhood defined by building machines to ask nature questions. But the cultural context made this particular build irresistible. A few years earlier, Cleve Backster — one of the country's leading polygraph examiners — had connected a galvanic skin response meter to a Dracaena leaf while watering it and claimed the plant's response pattern resembled that of a human subject experiencing emotional stimulation. When Backster merely thought about burning the leaf, the pen recorder reportedly went wild. He published his findings as "Evidence of a Primary Perception in Plant Life" in the International Journal of Parapsychology in 1968, and the idea that plants could feel, communicate, and even read human thoughts became one of the great pop-science memes of the early seventies. It would soon blossom into Peter Tompkins and Christopher Bird's bestseller The Secret Life of Plants in 1973, complete with claims about plants responding to music, prayer, and the emotional states of their owners from a thousand miles away.

Lawrence's article rode that wave shamelessly. A boxed sidebar titled "THE BACKSTER EFFECT" presented the claims without significant qualification. But Lawrence was also an electronics man, and to his credit, the instrument he designed was genuinely capable. He noted that plants were "organic semiconductors having variable resistance and self-generating properties," that typical electrical signals were in the low millivolt/microampere range, and that living matter tends to saturate and gradually cease functioning as an organic resistor, which is why he included the input/output polarizer switch to reverse current flow through the leaf. He warned builders to watch for power-line noise, RF interference, and the importance of controlled conditions. Whatever mysticism surrounded the project, the engineering was sound.

So I built one. I followed Lawrence's plans, assembled the Wheatstone bridge and op amp on a perf board, and mounted the controls on a metal chassis — and here the story owes everything to my father. He was the electrical engineer whose soldering iron had burned my curious fingers years before, and among his professional specialties was repairing Esterline-Angus strip chart recorders, the industrial-grade instruments used in laboratories and manufacturing plants to make continuous pen recordings of voltage, temperature, and process variables. He set me up with a very capable model, which meant that a high school kid with a homemade psychogalvanometer had access to the kind of precision recording equipment that most garage experimenters could only envy. I grew tomato plants in a hydroponic setup — I already had a fascination with controlled growing environments — and I set about testing whether my tomatoes had feelings.

They did not.

I played them music. I talked to them. I threatened them. I tried the full Backster protocol. The chart recorder traced flat, indifferent lines. My tomatoes were unmoved by Mozart, unbothered by menace, and supremely uninterested in my emotional state. If there was a primary perception in plant life, my plants had not received the memo.

A lesser scientist — or rather, a kid who wanted to win a science fair with a flashy title — might have stopped there, declared the experiment a failure, and moved on. But something Lawrence had written stuck with me. He'd noted that the resistance between the electrodes should not exceed 250,000 ohms, and that the bridge values had to be adjusted depending on the type of leaf being used because different plants presented different electrical characteristics. The instrument was measuring something real — the question was what.

So I changed the experiment. Instead of varying the stimulus while keeping the growing conditions constant, I kept the stimulus constant and varied the nutrient chemistry in my hydroponic solution. I set up four tanks of tomato plants: a control, and three treatments — calcium nitrate, ammonium phosphate monobasic, and potassium nitrate. Three different compounds, each delivering nitrogen through a different cation pathway, each carrying different ionic charges that would interact differently with the electrochemistry of plant cell membranes. I connected the psychogalvanometer to each set of plants in turn and recorded the response on the Esterline-Angus.

The results were unmistakable. The Esterline-Angus came alive. Different nutrient compounds produced significantly different readings across the leaf electrodes. What the psychogalvanometer was detecting was not plant emotion — it was ion exchange. The electrochemical gradient across the leaf tissue varied with the specific ions available in the root zone. Change the cation — calcium versus ammonium versus potassium — and you changed the electrical conductivity of the leaf. The plant was not feeling; it was signaling — broadcasting its physiological state through the same electrochemistry that Lawrence's instrument was designed to detect, if you knew what question to ask.

I did not, at the time, fully appreciate what I had stumbled onto. I was a teenager with a homemade galvanometer, my father's Esterline-Angus, and a hydroponic rack. The science fair judges were more intrigued by the apparatus than by the implications of the data. The project went into the archive of adolescent accomplishment, and I moved on to the next experiment — there was always a next experiment.

At Cal Poly, I persuaded my biology professors to give me lab access as a freshman and built machines to study the electric fields generated by Venezuelan knife fishes, devices to measure the rate of sap movement inside living trees, and an electronic hummingbird feeder that identified species by calculating wing-beat frequency. Every weekend I commuted to the San Jacinto Mountains for my job as a wilderness park aide. My friends considered me a serious example of the "early man" stage of nerd evolution, as happy backpacking through the wilderness as I was dissecting circuit boards. Eventually I landed at Cornell for a doctorate, spent thirty-six years directing UC field stations, and built a career on the premise that if you instrument the natural world carefully enough, it will tell you what it needs. But I never forgot the tomatoes.

Fifty-five years passed.

Then, on February 13, 2026, the Cornell Chronicle reported that the Center for Research on Programmable Plant Systems — CROPPS — had achieved a closed-loop system in which plants sense their own nitrogen status and automatically trigger nutrient delivery to correct it. The system is built around a genetically engineered tomato called RedAlert, whose leaves change color as nitrogen levels fluctuate, shifting from healthy green to red under nitrogen stress. Robotic cameras monitor the plants, custom imaging software quantifies the color shift in real time, and an automated decision-making pipeline triggers a microcontroller to adjust nitrogen delivery through the fertigation system. As nitrogen levels are restored, the plants return to green, and the cycle restarts.

Tomato plants. Signaling their nutrient status. Closing the loop.

Jacob Belding, a senior CROPPS trainee, framed the problem with precision that would have made my high school self weep with recognition: a huge percentage of the fertilizer American farmers apply — estimates range from forty to eighty percent — is never taken up by the plants. The remaining nitrogen becomes runoff, poisoning groundwater and surface waterways. Without a better source of information on the real-time needs of crops, we are forced to continue using nitrogen wastefully. The RedAlert system was built to change this by providing that crucial information.

The engineering is vastly more sophisticated than Lawrence's Wheatstone bridge. Where I had electrodes clamped to leaves with conductive jelly, CROPPS has genetically encoded reporter proteins that produce visible color change. Where I had a chart recorder tracing millivolt fluctuations, they have machine vision analyzing spatial and color data from an overhead camera gantry. Where I had a teenage experimenter adjusting fertilizer by hand, they have a programmable microcontroller actuating pumps and valves in a hydroponic fertigation system. The sprint that produced this demonstration brought together biologists, engineers, and computer scientists from multiple institutions — the kind of convergence research that didn't exist as a concept when I was reading Popular Electronics by flashlight.

But the core insight is identical. The measurable signal in plants is not mystical. It is electrochemical, or in the case of RedAlert, biochemical. The plant is reporting its physiological condition through mechanisms that can be detected, quantified, and acted upon. The question was never whether plants communicate — of course they do, through ion gradients, hormone signaling, volatile organic compounds, and a dozen other channels that modern plant science has catalogued. The question was whether we could build systems that listen to what plants are actually saying, rather than projecting what we wish they were saying.

Backster heard what he wanted to hear. Lawrence built an instrument capable of hearing the truth but framed it in the language of the paranormal. A high school kid in the early seventies, armed with a soldering iron, his father's strip chart recorder, and four hydroponic tanks of tomato plants, accidentally asked the right question and got a real answer: it's the ions, not the emotions.

And now, more than half a century later, a team at my alma mater has closed the loop. The plant speaks, the machine listens, the system responds, and the cycle begins again. From psychogalvanometer to programmable plant, the arc is not one of steadily advancing technology — though the technology has certainly advanced. It is an arc of learning to hear what living systems are actually telling us, which requires first letting go of what we hoped they would say.

I still have the photo of my build. The chassis with its gain and null knobs looks remarkably like Lawrence's prototype on page 67 of that June 1971 issue — same basic layout, same form factor, same ambition to make the invisible visible. The Esterline-Angus is long gone. The tomato plants returned to compost decades ago. But the question they helped me ask has turned out to be the right one, and it took fifty-five years and a team of brilliant Cornell engineers to give it a definitive answer.

Some loops just take a while to close.