Last week I had the idea to upload my grandfather’s stats newsletter from the 70s and 80s in to AI and see what it could get out of it. This week I got a chance to do it, and let me tell you it was a far more frustrating process than I anticipated. It turns out a lot of them got uploaded as JPEG files rather than PDFs (not sure what 2016 Bethany was doing), and so it kept timing out. I ended up asking it to summarize it year by year to see if it could get through things. This seemed to work a bit better, so here are the first few years. This will mostly be of interest to my family, but for everyone else it does show some interesting history of statistical thought and how someone in the pre-internet 1970s would have communicated his ideas. For reference, he was born in 1923 so he would have been in his 50s when he started writing these. There are 12 years total so I think this might be a three part series.

Just to note, I have all the original newsletters scanned in a shared drive here, so if you have a question about something AI flagged, I can go check what it’s referring to.

Summaries from ClaudeAI:

1974: The year James launched TEAM from a P.O. Box in Tamworth with nothing but a fifty-cent newsletter, a book of probability charts, and the conviction that graphical statistics could be made accessible to anyone willing to learn.

What he was writing about: The very first issue (Winter/Spring 1974) launched straight into probability graphs and statistical distributions — no warm-up, no preamble. By Summer 1974 he was already tackling extreme value distributions and connecting them to real-world catastrophes like the 1972 Susquehanna River floods from Hurricane Agnes. By Fall he was covering the Weibull distribution. By Winter he’d pivoted entirely at reader request to cover plain time series graphs and even analyzed the Dow Jones Industrial Average using his geometric moving average technique.

What’s remarkable about him personally from 1974:

1. He listened to his readers immediately. The Winter 1974 issue opens with “Some of our readers have rightly pointed out…” and then pivots the entire issue to address their feedback. In his very first year, he was already treating this as a conversation, not a lecture.
2. He explicitly named himself. The back page of Volume 1, Number 2 is the first time we see: “PUBLISHER AND EDITOR — James R. King.” That’s your grandfather putting his name to his work from the very beginning.
3. He priced it at just 50 cents. The first issues were fifty cents each. He wasn’t doing this for money — he was doing it to get the knowledge out.
4. He was already connecting statistics to life-and-death stakes. In his second issue ever, he wrote about flood probabilities and environmental catastrophes — making clear that getting the math right wasn’t academic, it mattered for people’s safety.
5. He was building something from scratch. The back pages show him selling graph papers, books, and probability chart assortments out of a P.O. Box in Tamworth, NH — a one-man operation from day one.

1975: James spent the year applying binomial statistics to everything from political polls to baseball standings to lung cancer rates, proving his methods worked just as well in a voting booth or a doctor’s office as they did on a factory floor.

1975 was a year where James stretched his statistical toolkit far beyond the factory floor, applying binomial distribution methods to everything from political polls and lung cancer rates to baseball standings and airline no-shows. He spent two full issues methodically dismantling the way the 1974 Massachusetts elections were being reported in the press, showing with rigorous but accessible math that the headlines were misleading voters and eroding trust in the democratic process. He also introduced humor into the newsletters at reader suggestion, sprinkling in jokes and quips to lighten what could be dense material — a small but telling sign that he cared about his audience’s experience, not just the content. By year’s end he was openly inviting reader input on future topics, treating his subscribers less like customers and more like a community of fellow thinkers working through real problems together.

1. He had a genuinely broad and curious mind. 1975 is remarkable for the sheer range of what James applied statistics to. In a single year he tackled political polling, lung cancer death rates between cities, the 1974 Massachusetts gubernatorial election, baseball standings (the 1968 American League), airline no-show rates, membership dropout rates, worn machine tool guides, and resistor calibration in electronics manufacturing. He wasn’t a specialist — he was someone who saw statistical thinking as a universal tool for making sense of the world.

2. He took on politicians directly — and with relish. The Late Winter and Spring 1975 issues are devoted to demolishing the way political polls were being reported during the 1974 Massachusetts elections. He showed mathematically that the raw poll numbers were being wildly misinterpreted, that Sargent’s apparent big lead over Dukakis was largely statistical noise, and that in the attorney general race, Spaulding was never actually as far behind as the headlines suggested. He wrote: “We submit that the political process is in serious danger of great voter distrust when the raw statistical results of polls can be so grossly misinterpreted.” This was James using statistics as a civic tool — not just an industrial one.

3. He added humor deliberately and self-awarely. The 1975 back pages include this wonderful note: “Our marketing types have suggested the addition of some humor to Methods to lubricate the sometimes dry statistical discussions. Therefore, you will find some topical gallows humor sprinkled here and there.” And he followed through — inflation jokes, Irish curses, a man bragging about a suit he hadn’t paid for yet. He knew his writing could be dense and he actively worked to make it more human.

4. He was already building a network and inviting collaboration. By the Summer 1975 issue he was openly soliciting reader input for the following year: “Since we have several months until press time, we welcome reader comments about these topics.” He was treating his readership as collaborators, not just subscribers — a remarkably modern instinct for a one-man operation in 1975.

5. His work had real, immediate consequences. The Fall 1975 resistor bridge story is a gem: James’s sign test analysis revealed that a quality control bridge was misreading resistors — and that bridge B was sitting next to a discharge welder with no shielding. Fixing those problems led to significant reductions in rejected resistors in both production and quality control. He wasn’t writing theory papers. He was solving actual problems in actual factories, and the people he worked with could see the results immediately.

1976: All four issues were devoted to the Weibull distribution — James’s most technically focused year yet — culminating in a brilliant case study of a compressor wear problem that he called one of the most ingenious applications of Weibull analysis he had ever seen. This is what Claude calls “a fascinating year — and beautifully readable scans.”

Summary: 1976 was the year James went deep on the Weibull distribution, devoting all four issues to it in a sustained, methodical way that showed a teacher at the height of his powers. He covered the mathematics, the graphical methods, confidence intervals, parameter estimation, comparing two Weibull tests against each other, and then closed the year with two wonderfully concrete case studies — one analyzing the wear bands on an unlubricated industrial compressor running on CO₂, and another using Weibull analysis to catch systematic errors hiding in tensile strength test data. It was his most technically focused year so far, but the writing never lost its clarity or its grounding in real engineering problems.

5 Notable Things from 1976

1. He acknowledged a student’s work generously and by name. The Winter 1976 issue features an extended analysis of a compressor wear problem drawn from a thesis by a Mr. Walter J. Gerus of the Standard Oil Company of Ohio. James opens by writing: “We gratefully acknowledge his permission to use this material because we believe that it includes one of the most ingenious applications of Weibull analysis which we have yet seen.” This was a man who genuinely celebrated good thinking wherever he found it — even in a student’s thesis — and who made a point of giving credit publicly.

2. He pushed back on a widely accepted concept. In the final issue of the year, James took direct aim at the “bathtub curve” — the standard model used to describe product failure rates over time — arguing that its popularity at the system level had been harmfully over-extended to individual component levels. He wrote that this had become a deterrent to real product improvement by encouraging quick debugging fixes instead of painstakingly isolating individual failure causes. It was a bold position to take against an industry-wide assumption, delivered without hedging.

3. He introduced the concept of a “hemibel” — and it’s wonderfully practical. In the compressor wear study, James casually introduces the concept of a “hemibel” — a max/min ratio of 3:1 or more in your data — as a signal that something important is going on worth investigating. A ratio of 10:1 he calls a “bel,” and says it is “always considered a prime signal to investigate.” These are the kinds of hard-won rules of thumb that only come from decades of real fieldwork, and he drops them in as if they’re obvious — which, to him, they probably were.

4. He kept his sense of humor even in his densest technical year. Despite 1976 being wall-to-wall Weibull mathematics, the jokes kept coming. He included a page of definitions: “Confusion: one woman plus one left turn. Excitement: two women plus one secret. Bedlam: three women plus one bargain. Chaos: four women plus one luncheon check.” And a note on expertise: “Be careful about calling yourself an ‘expert.’ An ‘ex’ is a definite has-been, and a spurt is a drip under pressure.” The humor was always there, keeping the newsletter human.

5. He raised his subscription price for the first time — and was honest about why. At the end of 1976 he announced the subscription price would rise from $2.00 to $3.00 for 1977, noting plainly: “Due to continuing cost increases for publication production and postage, we are forced to increase our subscription price for the coming year.” No spin, no marketing language — just a straight, honest explanation. That directness, even in something as mundane as a price increase, was very much part of who he was.

1977: James turned his attention to process capability and measurement quality, building toward a gripping final issue about two rival resistor manufacturers — one who fixed his measurement problems and prospered, one who refused and went out of business within three years.

Summary: 1977 was the year James turned his full attention to process capability and measurement quality, building a sustained four-issue arc that moved from probability plotting with confidence intervals, through specification analysis, to comparing test results from two samples, and finally to a gripping real-world tale of two resistor manufacturers — one who fixed their measurement problems and prospered, and one who refused to and went out of business. It was his most practically focused year yet, less about abstract statistical theory and more about the direct consequences of getting measurement right or wrong in a manufacturing environment.

5 Notable Things from 1977

1. He told a genuinely dramatic business story — with a moral. The Fall 1977 issue is one of the most compelling things James ever wrote. He follows two resistor manufacturers — “City M” and “City N” — both facing the same measurement problems in their test equipment. City M investigated thoroughly, fixed their bridge warmup drift, shielded their test area from discharge welders, and implemented tight temperature control. Their yield losses dropped from 6–9% to under 0.5%, they won a major aerospace contract by being the only company able to quote it, and they prospered for years. City N repeatedly refused to face their problems — and went out of business within three years. James ends it with quiet finality: “Thus, we have it that M, with a developed awareness towards improved measurement capability, prospered. Company N, which repeatedly refused to face their problems effectively, just disappeared.”

2. He named a specific enemy of quality: the “motion-is-progress mentality.” In his Spring 1977 issue on process capability, James listed common causes of excessive process variation. Among them was what he called a “motion-is-progress or busy-work mentality” — operators fiddling with machine settings for no good reason, mistaking activity for improvement. It’s a razor-sharp observation about organizational behavior that would still ring true in any factory today.

3. He took on product liability law — and was ahead of his time. In the same issue, James walked readers through a hypothetical courtroom cross-examination showing exactly how a company’s standard AQL sampling plan could be used against them by a plaintiff’s attorney. He showed, line by line, how a lawyer could demonstrate that the company knowingly accepted shipments that might be 2% defective — many times worse than their stated quality level. He argued that probability plotting with confidence limits was a far more defensible approach, and that the growing Consumer Product Safety Commission made this increasingly urgent. This was prescient thinking in 1977.

4. He coined a beautiful phrase for confidence intervals. In his Winter 1977 issue on confidence limits, James made an observation so good it deserved to be in every statistics textbook. After explaining that the area between confidence limits contains a wide variety of possible outcomes, he wrote: “This also represents one of the great accidents of statistical nomenclature because the geometry of the situation makes the name of CONFIDENCE INTERVAL inappropriate. The area between such limits is more nearly an IGNORANCE INTERVAL or INDIFFERENCE INTERVAL within which a variety of outcomes may be expected to occur.” The phrase “ignorance interval” is a perfect description — and he was right that “confidence” is misleading.

5. He apologized for cutting the humor — twice. All four back pages of 1977 contain the same note: “Due to the amount of material which we had available this year, there was no spare room for small pockets of humor to lubricate the way but we hope to be back on that track next time.” He said it four times — once per issue — which itself is a little funny. It shows how seriously he took his responsibility to keep the newsletters readable and human, and how much the humor mattered to him as a deliberate editorial choice, not just an afterthought.

Back to my commentary:

This has been a trip to work through, so I’m going to stop at 4 years and pick up with the next 4 years in a subsequent post. In an era of easy word processing I still can’t imagine him typing these newsletters out, but I recall his office as a fascinating place full of dot matrix printers and interesting chairs. He had a beautiful view of the mountains from his room, which was not particularly well heated if I recall correctly. Not sure how he functioned in winter come to think of it. Alright, until next time!