Domhoff, G. William (1996). Finding Meaning In Dreams: A Quantitative Approach. New York: Plenum.
 • Introduction
 • Chapter 1: The Scientific Study of Dream Content
 • Chapter 2: The Hall/Van De Castle System
 • Chapter 3: The Quality of the Data
 • Chapter 4: Normative Findings
 • Chapter 5: Age Differences in Dream Reports
 • Chapter 6: Cross-Cultural Studies of Dream Content
 • Chapter 7: Consistency and Change in Long Dream Series
 • Chapter 8: The Continuity Between Dreams and Waking Life
 • Chapter 9: The Repetition Dimension
 • References

Chapter 7: Consistency and Change in Long Dream Series


The previous chapters have established the potential of our coding system in terms of its reliability and its ability to yield findings compatible with common understandings of age, gender, and cultural differences. We are not surprised by the ways children's dream reports differ from those of adults, or men's from women's, nor by the cultural differences in some content categories. In a certain sense we are not enlightened by these findings, for they tell us nothing new about children, gender differences, or cross-cultural psychology. However, such findings give us confidence that our coding system is picking up psychologically relevant findings, and therefore may lead to new and interesting findings in future studies.

Previous chapters also have given us a normative context for studying the dream reports of specific individuals or groups of individuals. Because we know what is typical and atypical in the dreams of college students in general, we are in a position to understand the possible significance of unusual findings on individual college students or other population groups.

In this chapter we will provide another strong foundation for studying the dream reports of specific individuals by focusing on consistencies and changes in 100 or more dreams written down over months or years, which we call a long dream series. These dream series range in length from 187 to many hundreds of dream reports. The major finding emerging from the analysis is considerable consistency of dream content in categories with large frequencies. However, there are sometimes changes in some content categories in each series that make sense in terms of the person's altered life circumstances.

Although the primary focus of the chapter is on consistency, some of the findings will be compared with the Hall/Van de Castle norms in order to show the usefulness of group norms in understanding an individual dream series. By the end of the chapter we will have introduced two baselines for studying the dream reports of an individual--comparing the series with the norms, and comparing one subset of dreams within the series to another subset.

The several dream series to be analyzed in this chapter came to Hall from a variety of sources, but mostly from people who heard about his research and wrote to ask if their dream series (or "diary" or "journal" or "log," as they usually called it) could be of any use. Those who volunteered to send their series were asked to put a pseudonym on it and to type the dream reports if possible. Most people sent typed reports.

Far more people keep dream journals than may be realized. Such dream diarists come from all walks of life and keep journals for diverse reasons. Many people write down dreams as part of a more general personal diary. Some write down dreams as a possible source of short stories or paintings. Fewer than might be expected write them down for introspective psychological use, although the patients of Jungian analysts are an untapped treasure trove of long dream series. One house painter who sent Hall hundreds of dream reports originally wrote them down as part of his system for betting on horse racing. Whatever their original motivations, most dream diarists develop a routine that keeps them writing, and most say they rarely study their dream reports. Instead, the reports become a kind of record they value for its detail and continuity, and that they intend to look back at some time in the future.

There are three kinds of consistency in the results to be presented in this chapter--absolute, relative, and developmental. Absolute constancy means the frequency or percentage with which an element appears remains the same year after year. There will always be slight variations, of course, but they are not large, and there is no trend in them. Relative consistency means that the incidence of one element always exceeds the incidence of another element even though they both may increase or decrease in frequency; we will be giving examples of relative consistency among as many as four elements, but most of our measures of relative consistency concern two elements. Finally, there is developmental regularity, a consistent increase or decrease from one period of time to the next. The ways in which we measure each kind of consistency will be explained when we discuss the first dream series.

Relative consistency is the most frequent kind of consistency in dream series. Absolute constancy occurs slightly less often. Developmental regularities are much less common than the first two. This order of magnitude has held for each series analyzed to date.

We begin the chapter with a discussion of our longest and most carefully studied series. We present our methods and findings in detail with the hope that they can serve as a useful model for future studies, for such studies are very much needed. (Readers who are unlikely to do research on long dream series may want to skim many of the tables or ignore them entirely.) We then turn to briefer treatments of two other series quantified for three or more coding categories. Next we present findings on a long series that is unique because it was recorded over the space of a few months. Then we present findings on the consistency of the male/female percent in series from eight men and three women. We conclude with the results of a study of consistency in the laboratory (Kramer and Roth, 1979a).

Most of the analyses in this chapter were completed by Hall in the early 1960s before the Hall/Van de Castle coding system was finalized. Since there were few major changes from the earlier coding system except on the coding of friendliness and settings, the differences have very little impact, but they do mean that we cannot compare the findings to the norms on some content categories. In any case, the primary focus for now is on the phenomenon of consistency. We will have occasion in chapter 8 to use the norms with dream series analyzed more recently.

The Jason Series

Data and Method

We begin with the long dream series on which we have done the most extensive analyses. Jason kept a record of his dreams from age 37 until the time of his death in his early 70s. However, most of our analyses concern his 600 dream reports from between the ages of 37 and 54. Jason was a middle-class white professional. He was married and a parent when he started his dream journal. In his later years he left his wife, retired, and moved far from where he spent his work years.

Jason made it a point to write down every dream he remembered, except during a period of four years, when he only wrote down 15 dreams. He used no special techniques to recall dreams, nor did he make any particular effort to recall them. He was not a frequent recaller. He wrote dreams down exactly as he remembered them. The number of dreams he wrote down in each year of the 17-year span we will be analyzing can be found in Table 7.1.

Table 7.1. Jason's Dreams by Year and Age
  Year    # of dreams    Jason's age  

Jason kept his dreams for his own curiosity, but he had a detached attitude toward them. He recorded them as soon as he remembered them and then put them in a file. Every several months he would type them up and read through them.

The consistency scores on Jason's dream reports are presented in sets of 100 because the first important finding from this and other series is that it takes about 100 dream reports to establish stability from set to set on most content categories. For some categories, such as major character categories, there can be stability in subsets of 25 dream narratives, but for other categories it takes more, especially low-frequency categories. When we say there is consistency, then, we do not mean over a few dreams or over a few weeks. We are talking about hundreds of dreams and time spans of months or years. What people dream about usually varies from day to day and week to week. That variation, combined with the fact that most recalled dreams are soon forgotten, contributes to the conception of dreams as an aimless concoction of jumble and jive.[1]

We now need to explain the ways in which the three types of consistency were quantified in this and other long dream series. First, to measure absolute constancy, Hall decided to adopt a very simple statistic based on the average of the deviations of the sets from the overall average for all the dream reports together. He called this statistic v for "variation." The following example shows how it works, using the male/female percent from Jason's dream reports:

Table 7.2. Jason's Male/Female Percent over 600 Dream Reports
Dream #s 001-100  101-200  201-300  301-400  401-500  501-600  001-600 
Male/Female Percent63/3761/3957/4363/3762/3860/4061/39

The deviations of the six samples from the overall average are .02, 0, -.04, .02, .01, and -.01, respectively. The sum of these deviations, disregarding plus and minus signs, is .10. Dividing this figure by six yields an average deviation of .017 (.10/6 = .017). We then figure an average deviation by dividing by the smaller of the two percentages in the male/female percent. (We always use the smaller of the two percents in any calculation of v because it gives the most conservative estimate. v is far from being a sophisticated statistic because it becomes smaller as the percentage becomes higher. It is only a rough estimate.) To return to the example we are presenting here, the average deviation of .017 is then divided by the overall average for the 600 dream reports on female percent (39) to obtain v (v = .017/.39 = .04).

Now comes the arbitrary part. What is "excellent" or "good" or "poor" consistency? Hall defined them as follows:

  1. If the average deviation (v) is one tenth (10) or less of the percentage for the total population of dream reports, the consistency is said to be excellent.

  2. If v is between one tenth and two tenths (11 to 20) of the percentage for the total population of dream reports, the consistency is said to be good.

  3. If v is between two tenths and three tenths (21 to 30) of the percentage for the total population of dream reports, the consistency is said to be fair.

  4. Anything beyond 30 is considered poor consistency.

Since v is .04 for Jason's male/female percent in Table 7.2, the absolute constancy for this category is considered excellent.

Turning to relative consistency, defined as the stability with which codes for one category are consistently higher or lower than codes for another category, we will use two measures. When we are comparing two categories over a series of samples, it will be an expansion of the binomial theorem. When there is a large enough number of categories, we will use the rank order correlation (rho). The following example from Jason's dream series will demonstrate how the binomial theorem is used in determining the statistical significance of relative (also called row) consistency.

Table 7.3. Jason's Relative Consistency of Known to Unfamiliar Characters
Known Adult Males  30    32    25    22    19    27    26  
Unfamiliar Adult Males  17    15    24    25    25    21    21  

As can be seen, known adult males are more frequent than unfamiliar males in four of the six sets of dream reports. To determine the statistical significance of this finding, we have to expand the binomial theorem (1/2)6 to find the probability of obtaining a succession of six codes each of which is higher or lower than the other succession of six codes. The resulting probability is 1/64 or .016. Six out of six is statistically significant at the .016 level. Five out of six is significant at the .094 level (6/64), but four out of six could happen by chance 11 in 64 times, not very good odds that the difference is a non-chance one. In this case, we would reject the idea of relative consistency. So, we will use five or six differences in the same direction as our evidence for relative consistency in this study using six sets.

Developmental consistency can be seen in the following table showing Jason's A/C index with members of his family.

Table 7.4. Jason's A/C Index with Family Members
A/C with Family Members  .17    .19    .21    .25    .33    .68  

The table shows an increasing rate of aggressive encounters between the dreamer and members of his immediate family as he grows older. Once again using the binomial theorem, the criterion of developmental consistency will be five successive increments or decrements over the six samples of dream reports. In this case, then, there is developmental consistency for aggression with family members.

We are now ready for the overall findings. We will limit our discussion to characters and social interactions, but there is impressive consistency for our other categories as well.


Table 7.5 shows the total number of characters in Jason's dream reports, and the breakdown of those characters into ten classes.

Table 7.5. Frequencies and Percentages of Characters in Jason's Dreams Classified into Various Categories
Set: I II III IV V VI Overallv 
Total Characters  274  258  205  189  292  254  1472          
1. Single humans6365625559576008
2. Plural humans2021202020212002
3. Animals0201060806060544
4. Immediate family2124261917132018
5. Relatives01010000000100317
6. Known4441343431383811
7. Prominent0401010305050350
8. Strangers3033394447433914
9. Male Percent6361576362606104
10. Female Percent3739433738403904

In this table, absolute constancy, as measured by v, is extremely high for single humans and plural humans. It is low for animals, as it is likely to be with any content category appearing infrequently, given the statistical procedure being utilized. If the three categories (rows 1-3) are ranked, there is perfect relative consistency throughout the six sets of dream reports; the class of single humans is always highest, plural humans second, and animals lowest.

Turning to rows 4-8 of Table 7.5, we see no examples of good absolute constancy. The only important relative consistency is that family characters are always lower than either known or unfamiliar characters. As we already know because we used it as an example a moment ago, there is no relative consistency between known and unfamiliar male characters. The number of relatives and prominent persons in Jason's dream reports is too small to be of use in a study of relative consistency.

The percentage of male and female characters, as shown in rows nine and ten, is an example of both absolute constancy and relative consistency, but this finding comes as no surprise at this point in the book.

Seven character categories account for 87% of all the single characters appearing in Jason's dream reports. These seven categories are listed in Table 7.6 with their respective percentages for the six sets of dreams. The absolute constancy is excellent in the case of adult known females, good for adult known males and adult male strangers, and fair for person X, indefinite strangers, and adult female strangers. It is poor for person Y, who appears least often.

Table 7.6. The Seven Character Categories with the Highest Frequencies in Jason's Dreams
Character class  Dream numbers   v   
Adult known males3032252219272615
Adult male strangers1715242525212117
X (an individual)1113170812071223
Adult known females0907080910100909
Indefinite strangers0508071110060726
Adult female strangers0408060506110729
Y (an individual)0605080802030540
Numbers in table represent percentages of individual characters.

To summarize the findings on characters, a high degree of consistency as measured by v or by rank order of the categories (relative consistency) prevails among the various categories of characters in Jason's dream reports. There do not seem to be any systematic developmental regularities with respect to the characters appearing in Jason's dream reports between the ages of 37 and 54.


There are enough aggressions in Jason's dream reports to give us a very fair overall assessment of consistency. However, before presenting findings on consistency, we should note Jason has an extremely small amount of aggression in his dream reports. His overall A/C score is only 15 (198/1472), whereas the typical male dreamer has an A/C score of 34 (h = .45). We might want to allow for the fact Jason was in his middle 30s to early 50s when he was writing down these dream narratives, but he is indeed a very mild-mannered person, and he reports he always has been. Here we see the usefulness of our norms in studying individuals.

The data for aggressive encounters are brought together in Table 7.7. The first comparison aggregates all the aggressions in the dream reports. The next four comparisons are for aggressive interactions between Jason and single humans, plural humans, and animals, and for aggressions witnessed by him. These four categories constitute the sum total of all aggressions in Jason's dream reports.

The absolute row constancies (v) are good for the sum total of all aggressions and for Jason's aggressive encounters with single characters, and fair for witnessed aggressions. There are simply not enough aggressions involving plural characters and animals to draw any conclusions regarding consistency. The one developmental trend is the increasing proportion of aggressive encounters between Jason and single characters. In the last 100 dream reports, he is involved in almost twice as many aggressions with them than he is in the first 100 dream reports.

Table 7.7. Aggressive Encounters in Jason's Dreams (A/C Indexes)
 Number of
IIIIIIIVVVIOverall   v   
All characters198.
Single characters132.
Plural characters15.
Witnessed aggressions34.
Known characters33.
X (an individual)
Adult males30.
Adult known males20.
Adult known females11.
  IIIIIIIVVVIOverall   v   
Number of aggressions 292732384141198 
Victimization Percent 16%26%44%15%43%39%33%17
Physical Aggression Percent 00%09%22%13%32%09%15%10

In the rest of the rows in Table 7.7, the A/C scores are computed for single characters only, since it is with single characters that most of Jason's aggressions occur. First, we present the data for Jason's aggressive interactions with characters classified by their relationship to him (rows 6-9). We omit from consideration relatives and prominent people because there is only one aggression involving a relative and three involving a prominent person (there are only three relatives and 27 prominent persons in Jason's 600 dream reports). The data show clear-cut developmental consistency for aggression with family members. The consistency by rows for known persons and for strangers is just fair. There is no consistency in the relative ranks for known persons and strangers. The absolute constancy of Jason's aggressive encounters with males and females (rows 9 and 10) is good. Unlike most males, he has more aggressive interactions with females than males. His relative consistency on this issue is perfect.

We have selected four classes of characters for which there are sufficient aggressions to make an analysis of relative consistency feasible (rows 11-14). Although the changes for aggressions with individual X do not meet our criterion for developmental regularity, it is obvious that there is a much higher rate of aggressive interactions between Jason and individual X in the last 300 dream reports than in the first 300. His aggressive interactions with this person are invariably higher than with any of the other character classes.

The findings for victimization percent are presented in row 15 and the findings for physical aggression percentage are found in row 16. The absolute constancy on the victimization percent is good, and it is excellent on the physical aggression percent. Perhaps more interesting to readers becoming familiar with the norms is the fact that Jason is very atypical in the frequency with which he is the aggressor and in the small number of physical aggressions.

In short, the findings on aggressions not only show consistencies of various types for large-frequency categories, but they give us an image of Jason as well. He is not very aggressive, and not at all physically aggressive, but what aggression he has manifests itself through hostile interactions that he initiates with female characters.

Friendly and Sexual Interactions

As noted in the introduction to the chapter, this analysis was completed before the final Hall/Van de Castle system was settled upon. One of the major changes from the former system was the decision to separate out friendly and sexual interactions. Thus, the analyses in this section are not comparable with the norms. Taking out sexuality would entail a major reanalysis that would not change the consistency findings, but it would deflate scores. In a later part of this section on Jason's dream series, we will comment on the frequency of his sexual dream reports based on a greater span of years.

Table 7.8 brings together findings for Jason's friendly and sexual encounters with 14 classes of characters. In rows 1-5, the greatest number of these encounters is with single characters. The absolute constancy of rows 1 and 2 is good. The other categories have very low frequencies. There are no developmental trends in rows 1-5. The remaining rows (6-16) are based on an analysis of interactions with single characters. The stability of the scores is good for strangers, just fair for known persons, and poor for family members. The scores for known persons are higher than those for family members and strangers in every set of 100 dream reports.

Friendly encounters with males are invariably higher than those with females (rows 9 and 10), in contrast to Jason's greater rate of aggressions with females. Jason in fact manifests a very atypical pattern of friendly and aggressive interactions with male and female characters. The stability of the scores as measured by v is excellent for males and fair for females. There are no developmental regularities.

The rates of friendliness for the four classes of characters with whom Jason interacts in a friendly way most frequently appear in rows 11-14. The first two rows (11 and 12) are fairly stable, and the scores for adult known males outrank all other scores with a single exception. There is a suggestion of increasing friendliness with adult known females.

Friendly and sexual encounters are next divided into those in which the dreamer is the befriender and those in which he is befriended. Jason is fairly consistently the one who initiates the friendly or sexual encounter, as row 15 in Table 7.8 indicates, although the v is high due to one set of dream reports where the befriender percent deviates greatly from what is typical.

Table 7.8. Friendly Encounters in Jason's Dreams (F/C Indexes)
 Number of
IIIIIIIVVVIOverall   v   
All characters290.
Single characters249.
Plural characters13.
Witnessed friendliness15.
Known characters137.
Adult known males106.
Adult male strangers41.
Adult known females22.
Y (an individual)
  IIIIIIIVVVIOverall   v   
Number of friendly encounters 585136475048290 
Befriender Percent 83%85%75%89%72%53%76%40%

Overall, we think that a consideration of the quantitative tables on Jason's dream reports shows that there is an impressive amount of consistency over the 17-year period. Many of the v's are excellent, especially on general categories with high frequencies, such as males, females, and aggressions. Four of the poor (i.e., high) v's turn out to show an excellent degree of developmental regularity. Many of the remaining poor v's are high because of the low frequencies involved. Absolute constancy, then, is in the big picture, not the details. We also think the relative consistencies are impressive for several major categories.

A 30-Year Comparison

After a 15-year hiatus in receiving dream reports because we were doing other work, Jason gave us dream reports when he was 70 at our request. When we compared 100 of those dream reports with 100 from when he was 40, there were very few changes.

Throughout the first 21 years Jason kept track of his dreams, he had a large number of sexual dreams. At age 70, sexuality still remained an important element in Jason's dreams. He had a sexuality to characters index (S/C) of 15 (82/560) as a younger man, and it was about the same 30 years later. This norm is only 06 in male college students. Jason's higher rate of sexual encounters may mean there is an increase in sexual dreams for post-college men, but it is more likely the case that Jason had an unusual number of them.

There were three acts of physical aggression in the age-40 set, two in the age-70 set. The two in the second set were relatively mild and Jason said they caused him no discomfort: a large insect attacked him and a boy threw a stone at him.

We compared the two sets for Elements from the Past. Approximately one in ten dream reports can be coded on this scale in both sets. When it is taken into account that Jason had 30 more years of past life to dream about when he was 70, this result is all the more interesting. We will see further evidence for this "present-mindedness" of dreams with the next two subjects. We also see a similar phenomenon in a large cross-sectional questionnaire study of daydreams and other spontaneous thought intrusions in 1,275 men and women ages 17 to 92 (Giambra, 1977). The older subjects were no more likely to daydream about the past than the younger ones, and neither young nor old thought very much about the distant past.

The fact that Jason seldom dreams of the past also fits with the generalization emerging from studying the characters in his dream reports. He dreams about the people who are in his life at the time. The only overlaps of the two sets are his wife, offspring, father, a male friend he still saw, and his favorite college professor from 50 years earlier. Even when he was 40, all the known persons in his dream reports were current friends and associates except for his closest friend in high school, whom he had not seen in 20 years.

Given Jason's advanced age, we now looked for increased mentions of injury, illness, or death (misfortunes). There were four dreams of death and none of illness in the first set, and one of death and five of illness in the second set. In only one of these dreams, in the second set, was he the victim--an injury to his arm.

Jason died a few years after the second and final study. His contribution to our research was extraordinary. His dream series first raised the possibility that there may be consistency in what people dream about.


The longest dream series we have began in 1912 and ended in 1965, four days before the dreamer's death at the age of 78. During this 53-year span Dorothea wrote down 904 dreams. However, she did not start writing down every dream she remembered until 1959, well after she retired. There are only 54 dream reports from the ages 25 to 40 and 85 from ages 41 to 55. We therefore have to show a little caution about the early years when she only wrote down dreams that interested her, but in all it is a magnificent and useful series.

Dorothea was born in 1887 in Shanghai, China, where her parents were Presbyterian missionaries. She was the second of eight children. The family returned to the United States when Dorothea was 13. She taught school for many years after graduating from college. In 1925, she earned her doctorate in psychology and for the next 18 years taught in a normal school that later became part of a university. After her retirement at age 56, she remained active as a psychological researcher, publishing over a dozen articles in psychological journals. She never married.

Dorothea lived in a retirement home for women in a warm tropical climate for the last few years of her life. She was mentally alert and active until the very end of her life. She was a participating member of many local organizations. She belonged to a swimming club until she was 76. She was proud of the fact that she had an IQ score of 147 on the Wechsler when she was 74.

Although Dorothea was a psychologist, she had no professional or intellectual interest in dreams. She was an experimental psychologist with no use for Freud or other psychodynamic theorists. She wrote her dreams down for her own interest and did not contact Hall about her dream series until she was in her early 70s.

Hall studied 600 of the first 649 dream reports Dorothea sent him, dividing them into sets of 100. The first set covered her early adult and middle years. The last 100 in the study were written down when she was 75. The findings for Dorothea's series are consistent with those for Jason's, so we will not repeat the lengthy list of tables. The percentages of single characters, plural characters, and animals in her dream reports showed only minor fluctuations over the 50-year period. Nor were there any inconsistencies or high v's for the percentages of family members, friends and acquaintances, and strangers. However, there was a considerable change in her male/female percent in her later years, a developmental regularity. Her male/female percent was 53 in the first 100 dream reports, but only 39 in the 100 dreams from her 75th year. As the number of males in her life declined, so too did the number in her dream reports.

Dorothea dreamed about her parents and siblings with the same frequency throughout the fifty years, although her father died when she was very young and her mother died when Dorothea was 61. The number of aggressions, which were very low, did not change significantly over the years. Nor did the number of misfortunes.

There were only eight dream reports with sexual activity. Some of these dream reports are very explicit, so it does not seem likely she censored others from a dream journal she never showed to anyone until she was in her early 70s. At age 71, for example, she wrote down the following dream report:

As I stepped from the shower a man came in and said he wanted me. I said, "I'm all wet." He replied, "That's all right," and led me to the bedroom. I lay down on my back. Then he was on top of me. I felt his muscles moving and it felt like there was a rod up my vagina. I wondered if it was his penis inside.

Dorothea's comment that she "wondered if it was his penis inside" may reflect a genuine question. She wrote to Hall that she had never engaged in sexual intercourse.

In addition to the decline in her male/female percent, there were two other developmental regularities in Dorothea's dream reports. First, she dreamed more about traveling when she was a young woman, which is consistent with the fact that she did much traveling then and virtually none at all in her final five or six years. Second, she was higher on the misfortune of being ill in her earlier years, which corresponded with the fact that she was more troubled by illness when she was younger.

Madorah Smith and Hall (1964) did a separate study of the first 649 of Dorothea's dream reports to see if there was any tendency for her to dream more often of the past as she grew older. Using the Elements from the Past scale, they found no difference between 188 dream reports before age 66 and 461 reports after that age (23% vs. 29%). This finding is similar to what was reported for Jason's series in the previous section.

If Dorothea dreamt very little of aggression, sexuality, or the past, what then did she dream about? We will save part of the answer for a theoretical discussion of "themes" in long dream series in chapter 9, but the basic answer is that she dreamed most frequently about food and eating. Dorothea is eating, preparing to eat, preparing a meal, buying or seeing food, watching someone eat, or mentioning she is hungry in 128 of the 600 dream reports Hall used in the consistency study (21%), and in 85 of the 304 reports she sent later (28%), In many of these instances, she is at the family dinner table with her mother and some of her siblings. Her proportion of food and eating dreams is three times higher than our female norms.

Coincidentally, the last dream Dorothea mailed to Hall, four days before her death of old age, was an eating dream with her mother and some of her siblings in it. If we did not know how regularly she dreamt about being in this setting and, incidentally, not receiving her fair share of the food, we might infer that this final dream report was about an alleged forlorn existence in her retirement home or an omen of her impending death:

Mother had dished out too liberally to the younger children so asked E [a brother of Dorothea] to give her some of his. I still had nothing. Then we saw a potato on the floor by the door and it was divided with me.

It seems likely that this dream had to do with one of Dorothea's lifelong concerns, reminding us once again as to why norms or long dream series are essential for systematic studies of dream meaning.

We want to close this discussion by again underlining the fact that Dorothea was a very consistent dreamer for 50 years. As we said at the outset of this section, it would have been even better if we had more dream reports from her early adulthood, but the findings are still impressive. They become even more convincing after we consider the next long dream series.


Marie is one of the best coincidences that ever happened to quantitative content analysis. Between 1923 and 1932, when she was young and single, she recorded nearly 100 of her dreams. Most of them were written in 1923 or between 1926 and 1929. In 1933, she married and took a full-time job, and stopped writing down her dreams. Thirty-four years after her last dream entry, in 1966, Marie was a widow living in a small town in Central California several thousand miles from where she was born and raised. She started to write down her dreams again. Then she read Hall's The Meaning of Dreams (1953c) and wrote to ask if her dream reports would be of any interest to him.

When Hall quantified the dream reports, the major frequencies were virtually the same in both sets despite the gap of over three decades. This finding adds to what we already know from Jason and Dorothea because her early set of dream reports is concentrated in her early 20s. Moreover, her series is important because the continuity in content cannot be due to any "practice effect" from writing down her dreams continuously for the whole span of years. Thus, her series answers what was likely to be the major criticism psychologists would raise concerning our consistency findings.

Despite the quantitative similarities in the type of characters in the two sets of dream reports, the actual cast of characters changed almost completely except for members of her family of origin. Like Jason, she dreamed about people with whom she was in contact at the time, even if it was only by correspondence. Aside from her family of origin, the only other person she dreamed about in 1966-1967 who was not in any contact with her was her deceased husband. (She didn't know him when she first wrote down her dreams, so he does not appear in that set.)

For all the similarities in the two sets of reports, there were some changes as well. The percentage of unfamiliar characters is higher in the 1960s than the 1920s, for example. Although her rate of aggressive interactions is low in both sets, she dreamed more often of being a victim of aggression, especially from males, when she was in her 20s, whereas in her 60s she was more often the aggressor. Finally, the number of sex dreams was markedly less in her 60s. She reports she had been sexually active in the 1920s, but due to an operation in the early 1960s, along with the death of her husband, she no longer had sexual relations with anyone.

We can report that Marie was a wonderful person with a wide circle of friends and a great interest in working creatively with small hand tools. Her fascination with artistic objects came through in her dream reports, as when she was inspecting a rectangular piece of lucite exactly 2" x 1" x 1/2", or describing a beautiful Oriental rug, 8' x 10', in magnificent shades of red.

Marie's contribution to our growing realization that there may be consistency in dreams was very great. She is one of the reasons why we think the understanding of dream content will be advanced through the dream reports of normal adult citizens.

The Engine Man

The studies by Hall reported so far in this chapter establish consistency in dream reports over a period of years. However, there are not enough dreams in any of the series to tell us if consistency is possible over the space of a few months. A remarkable dream series obtained from a used book dealer by Alan Hobson (1988) provided an opportunity to study consistency in a shorter time period.

The person Hobson calls the "Engine Man" (because of his apparent fascination with train locomotives in his dreams) wrote down 234 dreams in the three months between July 14 and October 14 in 1939, strictly for his own curiosity. He was 46 years old at the time. When he died nearly 20 years later the dream journal was inherited by one of his relatives, who later sold it to the book dealer.

Hobson (1988) utilized these dreams to demonstrate the bizarreness and unreality of dreams in terms of their form, scene changes, and unlikely juxtapositions. Whatever the formal bizarreness of the dreams, they are extremely consistent in their content, as shown in an analysis by Adam Schneider of the 187 reports with 50 or more words. Table 7.9 presents these findings for several selected categories by comparing the overall findings with two subsets of 93 and four subsets of 46. The stability of the findings is not good with four subsets, but it is impressive with two subsets except for two or three categories, once again supporting our claim that 100 dream reports are needed to insure a representative sample.

Table 7.9. The Engine Man
    Total (187)    Two Sets (93/set)    Four Subsets (46 dreams per set)  
    Male/female %70707065766674
    Familiar characters %38383742344134
    Animal %11149151389
    Group %22251921291423
    Characters per dream3.
    Negative emotions %79847289777073
    Dreamer-inv. emotions %68686883468060
    Emotions per dream0.300.330.270.390.280.220.33
    Indoor settings %43454153374042
    Familiar settings %73767075766871
    D-inv. aggression %59516657456765
    Victimization %65626858676072
    Physical aggression %45424533503849
    Aggressions per dream0.550.460.620.460.480.460.80
    A/C index.
    D-inv. friendliness %70687271657073
    Befriender %52445750403868
    Friendliness per dream0.490.430.540.370.500.430.65
    F/C index.

The Engine Man was not only a consistent dreamer, but also a very unusual one, as revealed in his h-profile in Figure 7.1. He was below the norms in most aggression categories, particularly in terms of his involvement in aggressions. He is also strikingly below the norms in his involvement in friendliness. In other words, the Engine Man is in good part a witness or observer in his dreams. There is also a complete absence of sexual interactions from his dreams, not even so much as a kiss or sexual thought. He is above the norms only on animal percent and familiar settings.

Figure 7.1. h-Profile of "The Engine Man."
[Figure 7.1]

The Engine Man's h-profile anticipates findings we will present in the next chapter because it reveals a great continuity with his waking life, as presented in a four-page obituary in the proceedings of a scientific society. The Engine Man was a naturalist who did very detailed taxonomic studies, consistent with his high animal percent and his stance as an observer in his dreams. He was mild-mannered and unaggressive. He was a lifelong bachelor. We have no hesitancy in concluding that there is coherence, consistency and continuity in the dream life of the Engine Man. In fact, the findings with his dream journal suggest that consistency might be found in dream life over the space of several weeks for some people if enough dream reports could be obtained.

Consistency of the Male/Female Percent

As one small part of his study of the male/female percent in several dozen samples of male and female dream reports from all over the world, Hall (1984:1114) compared the male/female percent in early and later parts of 11 long dream series. Eight of the series were from men, three from women. As can be seen in Table 7.10, there is great consistency for periods ranging from a few months to many years. There is also consistency for Jason between samples 32 years apart. The only change is for Dorothea. Her male/female percent changed from 53/47 in 1912-1933 to 39/61 in 1960-1962, when she lived in a women's retirement home.

We report these results on the male/female percent because this is the largest sample we have relating to the issue of consistency. Male/female percent, since it has been of use in so many of our investigations, is a tell-tale indicator, but we recognize that more studies using more coding categories need to be done. We hope such studies will be completed and published by a variety of researchers in the next few years. Establishing consistency in a wide range of dream series would be of considerable theoretical import.

Table 7.10. Consistency Over Years for Individual Dreamers in Male/Female Percent
Individuals         Period 1         Period 2  
       Time interval    # of dreams    Male/Female %         Time interval    # of dreams    Male/Female %  
Karl 4/68-6/687669/31 7/68-9/687370/30
Tony 19685073/27 19695064/36
Ted 1966-196810072/28 1979-198010066/34
Dick 3/63-4/637461/39 5/63-6/637467/33
Chris 6/68-9/685068/32 9/68-12/685060/40
Raymond 19509161/39 19529162/38
Jasper 195910068/32 196310067/33
Jason 194710063/57 197912267/33
Lucille 19679942/58 19688643/57
Marie 19667951/49 19677948/52
Dorothea 1912-193310053/47 1960-1962100 39/61*
* p < .01

A Laboratory Study of Consistency

Kramer and Roth (1979a) have provided further support for the idea of consistency in dream content through an interesting laboratory study in which dream reports were collected from the first four REM periods for 20 consecutive nights from 14 college males. The 820 dream reports they obtained were analyzed for characters, activities, and descriptive elements using the Hall/Van de Castle system. A score for each of these content categories for each night was determined for each subject by adding the findings for all four dream reports of the night. Then the scores of all subjects for a given night were used to create a mean score for each night of the study.

Kramer and Roth used the nightly mean score to make a large number of comparisons for pairs of nights and for the first and second weeks of the study. Their most general measure was the mean night-to-night correlations for each of the three categories. They were .45 for total characters, .51 for total activities, and .51 for descriptions (Kramer and Roth, 1979a:322). After reporting that they found a correlation of .44 in their earlier study of the stability of sleep stages from night to night, they conclude that "the stability of dream content across nights is clearly comparable to that of sleep physiology" (Kramer and Roth, 1979a: 323). We think their conclusion is an impressive commentary on the consistency of dream content. Sleep stages are generally thought of as very regular and stable, just as dreams are thought of as irregular and unpredictable, but it turns out that the two different phenomena are similar in their consistency if dream reports are studied with quantitative content analysis.


The rather amazing consistency with which people dream about certain types of characters, social interactions, objects, and activities is a very strong argument for the idea that dreams have "meaning" in the sense of psychological coherence and significance. If dreams were only catch-as-catch-can reactions to internal physiological processes, or entirely at the mercy of too much salsa or a bad day at the office, it is unlikely that their contents would be so consistent over months or years or decades for this set of women and men.

The consistency in long dream series seems to be relatively free of fairly major changes in life circumstances if the findings on Jason, Dorothea, and Marie are any indication. They changed their geographical areas of residence and their family situations, and retired from their jobs, but they continued to have similar percentages and index scores for most dream content categories. At the same time, a few changes did occur. These changes are important because they help to refute any claim that the consistencies in our findings are a mere "practice effect" created by the process of writing down the dreams. The fact that the changes make sense in terms of changing circumstances in the dreamers' lives also strengthens our belief that our findings are psychologically meaningful.

In general, then, these findings add to our growing body of evidence that dream reports can tell us something meaningful about a person's life. They provide a starting point for future studies of long dream series from people who can give us more psychological information than is available on the people who provided the series used in this chapter. Indeed, we think the future of dream content studies lies in good part in the study of long dream series. If this chapter helps to legitimate such studies in the eyes of more social scientists, it will have gone a long way toward realizing one of the major goals of this book.

We think the findings in this chapter show that 100 dream reports are needed to establish reliable (repeatable) quantitative findings on a person's dream life. If we code sets of 25 dreams from a long dream series, we find more variation than if we code 50 or 75 or 100. This finding on reliability within sets from a long dream series is in effect another approach to the issues addressed in chapter 3 concerning the representativeness of a person's limited number of recalled dreams. In effect, these new findings from long dream series allow us to say that 100 dream reports are very likely to give us a representative sample of a person's total dream life. We think this rule of thumb is a good starting point for future "personality sketches" based on dream reports. These findings on the number of dream reports needed to ensure reliability in a dream series are the basis for the definition of a "long" dream series provided in chapter 1. A "long" dream series is defined as one containing 100 or more dream reports because it very likely contains a representative sample of a person's total dream life.

The finding that it takes 100 dream reports to obtain a representative sample of an individual's dream life is made all the more interesting by the fact that it dovetails with a comparable finding reported in chapter 4: it also takes 100 dream narratives to have a representative subset of the Hall/Van de Castle normative sample of 500 dream reports. Thus, for either individual or group studies, it is a mistake to generalize from studies containing less than 100 dream reports from each group or individual being studied.

Both a reliable coding system (one with high intercoder agreement) and repeatable findings are essential for the scientific study of dream content. However, the biggest challenge in terms of finding "meaning" in dreams is to link their content to waking thought and behavior, and to learn something new and important about dreamers that could not be learned through interviews and psychological testing. We turn to those challenges in chapter 8.

Continue on to Chapter 8


  1. Urbina and Grey (1975:360) determined "odd-even reliabilities" on the male/female percent in series only eight dreams in length, and found relatively low reliabilities, but their sample size was far too small to reject the idea of consistency, as we show in this chapter. [<<]

Go back to the Dream Library index.

dreamresearch.net home page dreamresearch.net contact info