James M. Norton, Ph.D.(1)
Department of Physiology
University of New England College of Osteopathic Medicine
11 Hill's Beach Road
Biddeford, ME 04005

Materials and Methods


Current understanding of the primary respiratory mechanism includes the concept of a linkage between the movements of the cranium and the sacrum via the spinal dura. To test this model for craniosacral interaction, methods were developed to document and analyze the timing of cranial and sacral cycles in healthy human subjects. A dual examiner protocol was utilized for a portion of this study, in which two examiners, one at the cranium and one at the sacrum, could simultaneously and independently document the cranial mechanism. A significant correlation was found between cranial and sacral cycle lengths documented separately by individual examiners, but pairs of examiners monitoring cranial and sacral cycle lengths of subjects simultaneously did not agree. These findings support predictions of the tissue pressure, or interactive, model for the cranial rhythm and do not support the concept of cranio-sacral interaction as described in the osteopathic literature.

key words: cranial rhythm; palpation; manual medicine; craniosacral therapy


Rhythmicity within the neural, respiratory, cardiovascular, endocrine and other systems plays a major role in the maintenance of stable and appropriate conditions within the internal environment of the body. One such rhythm is the "cranial rhythmic impulse" (CRI)(2) or "primary respiratory mechanism"(3), described as a rhythmic impulse arising within the cranium that is separate and distinct from any previously known pulsation and that is discernible on the external surface of the head with gentle palpation2. The movements of the cranium are thought to be linked with movements in the sacrum through mechanical forces transmitted through the spinal dura2. The resultant rhythmic activity of the sacrum is manifested by a slight rotation around a transverse axis slightly anterior to the second sacral segment1.

A recently published model for the palpation of the cranial rhythm(4) is based on the assumption that cardiovascular and respiratory rhythms and their contributions to soft tissue pressures of both subject and examiner are the primary determinants of the cranial rhythm as perceived during palpation. Experimental validation or refutation of this (or any other) model for the craniosacral rhythm requires data on frequency that are reproducible and accurate. Our laboratory developed an approach to studying the cranial mechanism of healthy human subjects that addresses these important issues, and began a series of experiments designed to test the interactive tissue pressure model and to assess inter-examiner agreement in the palpation of the cranial mechanism.

Preliminary findings recently reported by our laboratory indicate that the occurrence of rhythmic cranial cycles can indeed be directly documented by examiners(5),(6),(7),(8) using an unobtrusive, examiner-operated knee switch. Mean frequencies for healthy human subjects were found to be in the range of 3-6 cycles/min and examiner experience did not seem to affect the ability to palpate the frequency of the cranial mechanism. These early results suggested that the cranial mechanism's cycles could indeed be reliably documented, and that the basic timing of these cycles could be detected even by relatively inexperienced examiners.

If the cranial mechanism produces an independent and palpable physiological rhythm as has been suggested(9), then different examiners should be able to agree on the basic timing (cycle length and frequency) of the cranial rhythm in a given subject under a constant set of experimental conditions. Furthermore, if the linkage between cranial and sacral motion is real, then different examiners palpating the cranium and sacrum should document essentially the same frequency. In order to investigate this hypothesis, a dual examiner palpation protocol was developed in which subjects were monitored by two examiners simultaneously, one at the cranium and the other at the sacrum. This protocol was designed to investigate the presence of palpable cranio-sacral interaction as well as provide an indication of the extent of inter-examiner agreement.


subjects and examiners:

Subjects were students, faculty, or staff at the College of Osteopathic Medicine of the University of New England. The four men and five women ranged in age from 22-28 years, claimed to be in good general health, and were not suffering from any acute or chronic illness at the time of their participation in the study. All subjects participated voluntarily in the study, gave their informed consent to cranial palpation only (no treatments were given), and received no compensation. All of the 6 examiners were osteopathic physicians with extensive training and experience in cranial osteopathy.

general methods:

Silent switches activated by examiner knee pressure were attached to a leg and to the side of a fixed-height, wood-frame examining table. Examiners seated at the head of the table and at its side were asked to depress the switch at the beginning of the flexion (or inhalation) phase of the cranial mechanism and keep it depressed until the end of the flexion phase. For the cranium, the beginning of flexion was defined as the point at which the examiner felt a qualitative change in the direction of cranial movement toward "expansion"; the end of flexion was defined as the point at which the direction of movement reversed itself again. For the sacrum, the definitions were similar and were related to the rocking motion of the sacrum during the cycle. If the cranial mechanism is viewed as a succession of flexion and extension movements, then this protocol allowed documentation of the duration of one phase of the cycle (flexion or inhalation) and of the entire cycle (from the beginning of one flexion to the beginning of the next). The knee-switches were designed to provide simple, basic information about the timing of the cranial mechanism and was not intended to describe more complex information such as amplitude or the presence of restrictions or torsions.

A permanent record of the flexion phases was obtained using a chart recorder (Physioscribe, Stoelting, 620 Wheat Lane, Wood Dale, IL 60191). Activation of the knee switch produced an upward deflection in a time/event channel corresponding to the flexion phase of the craniosacral cycle. The paper speed used was 2 mm/sec in all measurement sessions; this speed was sufficient to allow direct measurements of flexion duration to the nearest 0.5 sec or estimates of flexion duration to the nearest 0.25 sec. Cycle lengths were routinely 15-20 sec in duration; measurement error was therefore in the range of 2.5-3.3%.

The basic data obtained for each measurement session using the protocol described above consisted of the duration of flexion (in seconds) and cranial cycle length (in seconds, from the beginning of one flexion to the beginning of the next) for 8-10 cycles. Cranial frequency (cycles/min) was calculated as 60 ÷ the mean cycle length in seconds. Calculated frequencies were validated by visually counting the number of flexions (or extensions) over a period of one minute on the permanent record.

For the dual examiner protocol, six [6] examiners trained and experienced in craniosacral techniques were paired and asked to monitor the cranial mechanism of subjects simultaneously, with one examiner at the cranium and the other at the sacrum. The two examiners were asked not to speak or otherwise communicate with one another and were prevented from picking up audible or visual cues from one another by a combination of soft background music and a large hanging curtain separating the examiners at the level of the subject's chest. During a measurement session, simultaneous records of flexion determined at the two sites were obtained using the separate knee switches for each examiner. Following a 1-2 min rest, the examiners switched positions and the measurements were repeated. Four subjects were monitored at the cranium and sacrum by all six examiners, and five other subjects were monitored at the cranium and sacrum by two examiners. A complete summary of the experimental data, expressed as frequency (cycles/min) is shown in Table 1. The protocol used allowed comparisons between cranial and sacral cycle lengths determined on a subject by the same examiner within a short period of time, and between cranial and sacral lengths determined simultaneously on a subject by two examiners.

statistical analysis:

Statistical analyses were performed using commercially available software (SigmaStat, Jandel Scientific, 65 Koch Road, Corte Madera, CA 94925). The tests utilized included one- and two-way analysis of variance, Pearson product moment correlation, Spearman rank order correlation, linear regression, Student-Newman-Keuls test for multiple comparisons, and Student's t-test. Sample size requirements for a power of 0.8, an of 0.05, and a mean difference in cycle length of 1 sec were met for the statistical tests used in this study.


A significant correlation was found between the cranial cycle lengths measured separately at the cranium and at the sacrum by the same examiner.

In the dual examiner protocol, the two examiners first palpated the cranial and sacral rhythms of a subject and then switched positions for a second set of measurements. Each examiner was therefore able to monitor the cranial mechanism of a subject at two separate sites, the cranium and the sacrum, within a very short period of time, usually less than 10 min. A highly significant correlation was found between cycle lengths (and therefore frequencies) documented separately by individual examiners at the cranium and at the sacrum of subjects, as shown in Figure 1 (top panel) and Table 2.

Agreement among examiners with respect to cranial and sacral cycle lengths was low.

Although cranial and sacral cycle lengths of subjects determined separately by individual examiners are correlated with one another, interexaminer agreement was low among the experienced examiners utilized in the dual examiner protocol with respect to cycle lengths measured at the cranium or the sacrum. Two-way analysis of variance of the dual examiner data revealed statistically significant degrees of variability in cranial cycle length attributable to both subjects (F = 4.54, p = 0.003, 7 degrees of freedom) and examiners (F = 20.006, p < 0.001, 5 degrees of freedom). These results indicate not only that the subjects differ from one another with respect to cranial cycle length but also that variability among the examiners in the documentation of cranial cycle length is greater than would be expected by chance even after accounting for the differences among the subjects.

When pairs of examiners document the cranial and sacral rhythms of a subject simultaneously, their findings do not agree.

In contrast to the highly significant correlation between the cranial and sacral cycle lengths of a subject determined separately by the same examiner, no significant correlation could be demonstrated between the cranial and sacral cycle lengths of a subject determined simultaneously by different examiners in the dual examiner protocol (Figure 1, bottom panel, and Table 2). In addition, as also shown in Table 2, no statistically significant correlations were found between cycle lengths measured separately at the cranium or the sacrum of a subject by the two examiners during a session. Furthermore, no consistent temporal relationship could be established between the onset of flexion as documented by the two examiners on the chart record of the experiment. There was no visual evidence of a time delay or phase shift (suggesting a fluid or mechanical wave moving caudally) or of one rhythm being a simple multiple or harmonic of the other.


Our protocol required the examiners to document their subjective palpatory findings in a manner that produced a permanent record for subsequent analysis. Spoken words, nods of the head, or other gestures used to indicate palpatory findings might be heard or observed by subjects and other examiners, and would require the assistance of an assistant to record the signals. We wanted the permanent record to represent as accurately as possible the examiners' actual perceptions, and chose the knee switch as the vehicle for making the most direct link between examiner perception and permanent documentation.

The knee-switch method used to collect the data described in this report did not appear to interfere at all with the interaction between the examiners and the subjects. Subjects were usually completely unaware of the means by which the recordings were obtained. The slight movement of the examiner's leg required to depress the switch was nearly imperceptible to a person lying on the treatment table. When questioned informally after the measurement sessions were completed, new examiners admitted that it took several cycles to get used to the switch, but from then on the experimental hardware and setup did not interfere with their ability to palpate the cranial mechanism of a subject. All examiners felt that the switch allowed them to record their palpatory findings directly and accurately.

Several control studies were performed to assess the ability of examiners to document the timing of cranial cycles using the knee switch. First, it was shown that examiners could use the knee switch to palpate and document accurately the timing of the inspiratory phase of the respiratory cycle, using subjects whose respirations were monitored and recorded using a pneumograph. Respiratory cycles were shorter than the cranial cycles observed in this study and therefore would theoretically have a larger measurement error using our system. Nevertheless, the correlation between respiratory cycle lengths recorded directly and those documented by an examiner was highly significant (Pearson's r = 0.934, p < 0.001).

In a second test of the measurement system, one examiner was asked to document the cranial cycles of three subjects by marking first flexion phases for 6-8 cycles and then extension phases for a similar number of cycles. Mean cycle lengths determined using the flexion and extension phases were 24.5 ± 4.5 sec and 25.2 ± 5.6 sec, respectively, with no significant difference between the two sets of measurements. Thus, regardless of the phase used by the examiner to document the timing of a subject's cranial mechanism, the mean cycle lengths (and therefore the frequencies) were the same.

Cranial cycles do not have to be regular in order for their timing to be documented accurately. Fourier analysis can be used to analyze signals with a varying periodicity(10), and this method can also be applied to time-domain signals that are square waves (periodic step functions) similar to the chart records produced in our protocol. In a third type of control experiment designed to assess our protocol, the cranial mechanism of a one subject was monitored by an experienced examiner for a period of more than 11 min, allowing documentation of 40 cycles with an average duration of 17.19 sec (corresponding to a frequency of 3.49 cycles/min). Fourier analysis of the record transcribed into a digital form (1 = flexion, 0 = no flexion) at 0.5 sec intervals (a total of 1,344 data points) produced a sharp peak in the frequency spectrum at 3.4 cycles/min. Such agreement supports the use of our knee-switch system to record digitally the timing of an analog process with the potential for varying periodicity.

The overall cycle length was chosen as the measurement for comparison because it was felt to be less sensitive to differences in examiner palpatory skills than the duration of flexion alone. The underlying assumption was that differences in examiner skill or experience might affect the point in the cycle at which flexion is perceived to begin, but that this point would be essentially the same from cycle to cycle. Since the beginning and ending of flexion is sensed as a qualitative change in the direction of movement, documentation of the timing of a subject's cranial mechanism by the experienced examiners utilized in this study should not be grossly affected by variations in the baseline amplitude of the mechanism itself. All examiners so questioned stated that the cranial rhythms of the subjects used in this study were readily palpable and no different than those regularly encountered in their practice.

Although cycle length is not usually discussed formally or informally as an important feature of the cranial mechanism, it is directly related to frequency, considered to be an important characteristic. As described above, the data analyses in this report were performed on original cycle length data rather than on calculated frequencies, in order to enhance statistical validity. Since the time and frequency domains simply represent different ways to describe a rhythmic process, conclusions drawn from cycle length can be readily applied to frequency, and vice versa. In support of the findings presented here, two recent investigations of the cranial rhythm in which the methods for documenting frequency are clearly described have yielded average frequencies in the same range as those described above(11),(12)

It is difficult for this author to reconcile the lack of interexaminer agreement seen in this study with the existence of an independent, easily palpable physiological rhythm generated within the cranium of a subject and transmitted to the sacrum, since simultaneous measurements of cycle length obtained on a subject at the two locations by different examiners were not correlated at all (Table 2). Furthermore, the results seem much more consistent with a rhythm that is only perceived by an examiner to come from a subject but that actually arises somehow from the interaction between subject and examiner. This statement is supported by the fact that measurements of cycle length obtained separately at the cranium and sacrum of a subject by the same examiner were significantly correlated with one another, but the cranial (or sacral) cycle lengths of a subject determined by multiple examiners did not agree. These findings are consistent with the interactive tissue pressure model3, according to which an examiner would be expected to perceive the same rhythm in the soft tissues of a subject regardless of the site of palpation. Different examiners would not be expected to perceive the same cranial rhythm on a given subject because the persons involved in the interaction are different and therefore the physiological rhythms combining to produce the palpated cranial rhythm would be different.

These data do not support the "membrane pulley model"1 or "spinal reciprocal tension membrane"2 hypotheses for craniosacral interaction, which would predict that movements or rhythms at the cranium would be causally and temporally related to movements at the sacrum, with respect to both mean frequency and the point of onset of the flexion phase(13),(14). We could demonstrate no significant correlation between cranial and sacral rhythms as palpated simultaneously by two examiners, either statistically or by visual inspection of the experimental records.

In conclusion, our results do not support the existence of a craniosacral rhythm that arises within a subject and that is capable of being consistently documented by experienced examiners. Several possible explanations of our data come to mind: 1) the cranial rhythm does not exist, as suggested in several recent publications on the subject(15),(16); 2) inter-examiner reliability in cranial palpation, an essential prerequisite for any meaningful clinical trials of the efficacy of craniosacral therapy(17), is very poor; and 3) the perception of motion arises from some aspect(s) of the interaction between an examiner and a subject, in which case interexaminer agreement would be expected to be low and the ability of practitioners of craniosacral therapy to share accurate and objective information would therefore be limited.

Craniosacral therapy continues to be practiced in various forms by physicians, physical therapists and others. Such widespread use of this modality demands further research to confirm the existence and nature of the cranial mechanism, to establish the reliability of information obtained during cranial palpation, to define the characteristics of a normal mechanism, and to generate criteria for determining the efficacy of craniosacral therapy.


The author would like to acknowledge the significant contributions to this study of Richard Broder-Oldach, D.O., and Gretchen Sibley, D.O., both students at the University of New England College of Osteopathic Medicine at the time the experiments were performed.

FIGURE 1 (top panel)

FIGURE 1 (lower panel)

Legend for Figure 1

Scatter plots of data obtained during the dual examiner experiments. Each point represents average values for cranial and sacral cycle lengths for a single measurement session. The upper panel compares cycle lengths from all subjects at the two locations documented separately by the same examiner; the lower panel compares cycle lengths documented by two different examiners simultaneously.

Table 1.

Dual Examiner Data Summary for Cranial and Sacral Frequencies(18),(19)



S1 S2 S3 S4 S5 S6 S7 S8 S9
E1 cranium 5.50 5.17 5.49 6.70
4.68 6.32 5.06 4.32
E2 sacrum 3.43 2.85 3.65 3.56
2.91 3.48 2.91 3.44
E1 sacrum 7.38 5.06 6.18 6.59
4.97 7.26 5.22 4.35
E2 cranium 3.35 2.86 4.44 3.90
2.96 3.79 3.57 3.35
E3 cranium

2.60 2.47 2.06 2.20
E4 sacrum

4.08 4.26 4.72 4.09
E3 sacrum

2.15 2.20 2.14 2.15
E4 cranium

4.06 6.13 5.53 4.61
E5 cranium

4.89 4.10 4.93 4.09 3.63
E6 sacrum

6.47 5.38 6.82 4.34 7.06
E5 sacrum

4.47 3.93 4.72 3.50 2.94
E6 cranium

5.58 5.23 7.35 4.78 5.71


Table 2.

Correlation(20)of Cranial and Sacral Cycle Lengths(21)

sacral(A)(22) cranial(B)c sacral(B)c
cranial(A)c 0.926 



1. James M. Norton, Ph.D., is Professor and Chairman of the Department of Physiology at the University of New England College of Osteopathic Medicine, 11 Hill's Beach Road, Biddeford, ME 04005. Address all correspondence and requests for reprints to Dr. Norton.  This study was approved by the Institutional Review Board of the University of New England, and was supported by grant #91-14-345 from the American Osteopathic Association.

2. Upledger JE, Vredevoogd JD. Craniosacral Therapy. Chicago, Eastland Press, 1983, p. 6.

3. Magoun, H.I. Osteopathy in the Cranial Field. Kirksville: the Journal Printing Company, 1976.

4. Norton JM. A tissue pressure model for the palpatory perception of the cranial rhythmic impulse. JAOA 1991;91(10):975-994.

5. Norton JM, Sibley G, Broder-Oldach R. Quantification of the cranial rhythmic impulse in human subjects. JAOA 1992;92(10):1285.

6. Sibley G, Norton JM, Broder-Oldach R. Interexaminer agreement in the characterization of the cranial rhythmic impulse. JAOA 1992;92(10):1285.

7. Norton JM, Sibley G, Broder-Oldach R. Characterization of the cranial rhythmic impulse in healthy human adults. AAO Journal 1992;2(3):9-12, 26.

8. Norton, J.M. Documentation of the cranial rhythmic impulse. STILL Alive 1(1):January 1996. (an electronic journal, URL is http://www.rscom.com/osteo/journal/stalive.htm.)

9. Upledger JE. The reproducibility of craniosacral findings: a statistical analysis. JAOA 1977;76:890-899.

10. Noggle JH. QuickBasic programming for scientists and engineers. Ann Arbor, MI; CRC press, 1992, pp. 301-310.

11. Hanten, WP. personal communication.

12. Wirth-Pattullo, V., and K.W. Hayes. Interrater reliability of craniosacral rate measurements and their relationship with subjects' and examiners' heart and respiratory rate measurements. PhysicalTherapy 1994;74(10)908-920.

13. Lee RP. Primary and secondary respiration. Part II. AAOJournal 1993;3(1):17-19, 27.

14. Lee RP. A report to the statutory advisory committee on medical care: craniosacral manipulation. AAO Journal 1991;1(1):13-14.

15. Wirth-Pattullo, V. Interexaminer reliability of palpating the craniosacral motion in the clinical setting. PhysTher 1993 73(6):S10.

16. Ferre JC, Barbin JY. The osteopathic cranial concept: fact or fiction? Surg Radiol Anat 1991 13:165-170.

17. Johnston WL. Interexaminer reliability studies: spanning a gap in medical research - Louisa Burns Memorial Lecture. JAOA 1982 81:819-29.

18. Frequencies are expressed as cycles/min and were calculated as 60 divided by the mean cycle length (in sec) determined for a subject by an examiner.

19. Subjects are designated as S1-S8, and examiners as E1-E6. Examiner pairings were E1 and E2, E3 and E4, E5 and E6. Data are grouped by session, with examiners switching positions between the cranium and the sacrum.

20. The Pearson product moment correlation was utilized to produce this table; the data given for each comparison are the correlation coefficient, the P value, and the number of samples compared.

21. Data shown is from thirty-four [34] measurement sessions utilizing the dual examiner protocol as described in the text.

22. The comparisons made in this table are: cranial(A) to sacral(A), pairs of measurements made separately at the cranium and sacrum of a subject by the same examiner during a measurement session; cranial(A) to cranial(B), pairs of measurements made at the cranium of a subject by two different examiners during a measurement session; cranial(A) to sacral(B), pairs of measurements made simultaneously on a subject by two different examiners during a measurement session; sacral(A) to sacral(B), pairs of measurements made at the sacrum of a subject by two different examiners during a measurement session.