lf.A personal publicationItaliano
Thresholdsliminalfinds.com

The Unsharpened Pencil

In 1893, Mary Whiton Calkins placed paper, candle, and pencil beside her bed to tabulate 375 dreams—and caught the instant a record loses what it meant to hold.

Graphite pencil drawing on pale paper of a bedside table from above: a completely blank sheet of paper beside a hexagonal wooden pencil with an uncut blunt end, an unlit candle stub in a saucer, and a closed matchbox.
AI-generated editorial illustration.

The apparatus was ordinary: paper, pencil, candle, and matches placed within arm's reach of the bed. On waking from a dream, the sleeper was to jot down every remembered detail before it faded, then read those brief notes the next morning, copy them out in full, and trace what had carried over from waking life. One night during the study, an experimenter woke from a dream "apparently of peculiar significance," wrote the whole account out in the dark, and went back to sleep "with the peaceful consciousness of a scientific duty well done." In the morning there was an unsharpened pencil beside a blank sheet of paper, and not the remotest memory of the dream.

Mary Whiton Calkins tells that story on the second page of "Statistics of Dreams," published in The American Journal of Psychology in April 1893. She signs the paper as instructor in psychology at Wellesley College, with a footnote recording her year as a private pupil in 1890–91 and noting that the study was undertaken at the suggestion of Dr. Sanford. The dated entries that carry weekdays—Friday, March 13–14; Tuesday, March 31; Thursday, May 7—fit the calendar of spring 1891.

Two observers kept the night logs for six to eight weeks: C., Calkins herself, then twenty-eight, who recorded 205 dreams over 55 nights, and S., a thirty-two-year-old man whose name is withheld, who recorded 170 dreams over 46 nights. Together they assembled 375 dreams, averaging nearly four a night, even though neither had previously considered themselves a frequent dreamer. Calkins does not say whether the unsharpened pencil belonged to her or to S.

An instrument that disturbs what it counts

Before settling on natural awakenings, they tried an alarm clock set for different hours of the night. They gave it up after the opening weeks because "the excitement of waking through its agency was often so great as to prevent connected memory of the dream." Even without a bell, the threshold between sleep and writing proved brittle. The mere motion of reaching for paper or lighting a candle could scatter the scene, leaving "the tantalizing consciousness of having lived through an interesting dream-experience of which one has not the faintest memory."

Nor did keeping the record leave sleep untouched. Calkins warns that an observer with recording "on his mind" may sleep less soundly, dream more, and wake oftener; and she adds that morning reflection easily supplies order that never existed in sleep, reading deliberation and emotion into a drift of images "something as one interprets an animal's actions by one's own." Her opening rule sets the ceiling for all twenty-one tables that follow: "At the best, one may discuss only dreams as remembered."

Looked at through that sieve, the time-of-night figures carry an explicit caveat. In C.'s 183 timed dreams, 134 (73.2%) occurred after 4 a.m. and 49 (26.8%) before 4 a.m.; in S.'s 118 timed dreams, 62 (52.6%) fell between 5:31 and 8:30 a.m., 52 (44%) between 11:31 p.m. and 5:30 a.m., and 4 (3.4%) between 10:30 and 11:30 p.m. Yet Calkins insists that middle-of-the-night sleep "is in no sense a dreamless sleep." Once the alarm clock was dropped, fewer night dreams entered the tables simply because the sleepers woke less often to write them down.

Writing the record inside the dream

Other patterns in the 375 records held up better against the bias of recall. Across the entire corpus, Calkins writes, "there is no instance of a dream in any other than the present time." Even when a dream returned to a childhood home or brought back someone unseen for years, "the apparent age of the dreamer was never lessened to avoid anachronism"—saving only the rare dreams in which personal identity itself shifted.

What entered those dreams was seldom what occupied waking thought most deeply. In Table XX, trivial episodes outnumber personal or professional ones: 104 of C.'s and 86 of S.'s general-character dreams fall into the trivial column (such as S. dreaming of a large picnic party boating on a pond), and Calkins notes that even the "personal" column—like C.'s dream of Mrs. S. at lunch with mismatched coffee cups—seldom touches "really important phases of personal relation." Calkins confirms Yves Delage's observation from July 1891 that in bereavement one rarely dreams of the dead. At least twice during the months of observation, after days filled with vivid memories of a friend who had died, an observer went to sleep expecting to dream of them and dreamed only of trivialities—though Calkins rejects Delage's explanation that waking attention burns up an image's supply of energy.

Among the false recognitions—paramnesia—that filled the notebooks, one recurring form stands beside the blank page from page 312. "A favorite type of paramnesia, especially with one of the experimenters," Calkins writes on page 337, "was the dream of writing the records of a previous dream." Read together, the two moments trace the edges of the bedside notebook from opposite sides: in the dark, a waking hand writes with an uncut stick of cedar and leaves blank paper; in sleep, the hand writes down a neat record that exists nowhere outside another dream.

The closing pages test the opposite temptation: treating an unlikely match as foresight. Five dreams in the two notebooks matched subsequent events, including C.'s Dream 19 (11:35 p.m. on March 13–14), in which she rides a Cambridge streetcar with a Miss M. in a red cloak. Two weeks later, on March 28, she met Miss M. on a Cambridge streetcar during a two-day visit home after more than a year in Brooklyn. Calkins refuses any telepathic inference, pointing to forgotten cues and sample size: "If in a large number of dream experiences, suggested by the affairs of daily life and peopled with the larger part of one's acquaintance, there should be no coincidences, that in itself would be a most remarkable fact."

Three years later, in April 1896, Wellesley students Sarah C. Weed and Florence M. Hallam, assisted by Emma D. Phinney and communicated by Calkins, published a 381-dream follow-up on the same bedside model. There too, the one participant who could only write down her dreams in the morning logged 50 dreams over five or six weeks, compared with 141 and 150 for the two who wrote immediately on waking. Calkins had already summed up what a notebook by the bed can and cannot settle: slow, ordinary self-observation will rarely yield startling discoveries, yet it keeps old generalizations honest against what actually remains on the page by morning.

02 / A Note on Sources

A Note on Sources

Read in the complete scanned PDF, JSTOR XML, and OCR text of Mary Whiton Calkins, "Statistics of Dreams," The American Journal of Psychology, Vol. 5 (April 1893), pp. 311–343, on the Internet Archive (jstor-1410996), with all quoted passages and Tables I–IV, XIX, and XX verified directly against rendered page images.

What is added. Calkins's 1893 paper is well known in the history of sleep psychology, and the unsharpened-pencil anecdote is occasionally cited as a curiosity. What we set side by side here—checked against the original printed pages—is the symmetry between the blank sheet written in the dark on page 312, the alarm clock abandoned on pages 311 and 314, and the recurring dream of writing down an earlier dream on page 337, alongside her statistical refusal of prophetic coincidence on pages 342–343.

Facts and reading. All counts (375 dreams total: 205 by C. over 55 nights, 170 by S. over 46 nights; Table I–II time splits; Table XX triviality counts), quotations, and the 1896 Wellesley follow-up counts (Weed, Hallam, and Phinney, 381 dreams, jstor-1411389) are taken directly from the two primary journal issues. Calendar alignment of Friday March 13, Tuesday March 31, and Thursday May 7 to spring 1891 was checked by standard calendar calculation. Reading the unsharpened pencil and the dream of recording a dream as two inverse failures of the bedside threshold is our interpretation.

What remains unknown. Calkins withholds the full identity of Observer S. (describing him only as a 32-year-old man), and she does not specify whether the unsharpened-pencil night or the recurring paramnesia of recording an earlier dream occurred to C. or to S. Neither the original handwritten bedside notebooks nor Sanford's correspondence around the 1891 study were consulted.

03 / The find

Mary Whiton Calkins, "Statistics of Dreams" (April 1893), pp. 311–343

A 375-dream bedside study from 1893 where an alarm clock broke recall, a dream written in the dark with an unsharpened pencil left a blank page, and one of the sleepers kept dreaming of writing down the record of an earlier dream.

The 1893 study (Internet Archive)

Wellesley follow-up (1896)

How this was written: researched and drafted with the help of AI tools, starting from the sources listed here. Editorial responsibility for what is published is mine. Corrections: hello@liminalfinds.com.

If this was worth your time, you can support Liminal Finds (opens in a new tab).