In several studies, neural-tissue preparations were exposed to RFR in vitro. Among them was that of Courtney et al. (1975), who removed the right and left superior cervical ganglia from New Zealand white rabbits, stretched each across a vertical section of WR284 waveguide sealed at the bottom and filled with Ringer's solution, and exposed the ganglion to 2.45-GHz CW RFR from below. Temperature-controlled Ringer's solution was flushed through the waveguide section at 1 l/min. For impedance matching, the seal at the bottom of the waveguide was a dielectric plate of unstated thickness (presumably equivalent to a quarter-wavelength). The preganglionic end was connected to a set of stimulating electrodes external to the waveguide, and a silver lead wire within a suction syringe (external to the waveguide) was used to make contact with the postganglionic nerve via a glass capillary through the waveguide wall.

Forward and reflected powers were measured with a bidirectional coupler and power meters. Reflected powers were always less than 1% of forward powers. The mean power density incident on the ganglion and its mean SAR were calculated from the net (forward minus reflected) power, the waveguide cross section, the penetration depth for the Ringer's solution (1.75 cm at 2.45 GHz), and the distance of the ganglion from the bottom of the waveguide, which yielded 2.2 W/kg per mW/sq cm.

Prior to exposure, effects of temperature alone on synaptic transmission latencies through ganglia were measured. At 37 deg C, stimulation of the preganglionic nerve trunk with pulses 100 to 300 microseconds in duration at 1 pps and amplitude 4-8 V yielded a transient positive-going potential from the postganglionic nerve stump relative to the potential of the Ringer's solution, with a transmission latency of about 20 ms (B-fiber-mediated or short-latency response). Seen with higher stimulus amplitudes were C-fiber-mediated (long-latency) responses. Latencies for both responses increased about linearly with decreasing solution temperatures, yielding at 23 deg C about 40 ms and 65 ms for the short-latency and long-latency responses, respectively, which corresponded to change rates of about 3% per deg C for both.

Each ganglion was exposed to the RFR at power densities up to about 300 mW/sq cm (660 W/kg) for 1-min intervals, with 1 min between exposures for control measurements. During the exposure and control intervals, the ganglion was stimulated with an amplitude sufficient to excite both B and C fibers and the electrically evoked potentials were recorded with a computer of average transients. The authors stated: "Only at absorbed power densities [SARs] above 100 W/kg (50 mW/sq cm) did we observe temperature changes of 0.1 deg C or more in the solution just exiting the waveguide." However, they did not discuss the method for measuring such temperature changes.

Averaged short-latency and long-latency responses were plotted vs RFR level for each of three ganglia. These plots showed small differences in latency between exposure and control periods for each ganglion. The authors gave no statistical treatment of the data but indicated that by t-test, such differences were nonsignificant (p>0.05) for either type of latency. The plots for two of these ganglia also showed temperature rises exceeding 0.2 and 0.4 deg C at the respective highest RFR levels.

The authors noted that some changes in response latencies over time were observed. They presented a plot of the short latencies at 36.5 deg C vs time for a ganglion in the absence of RFR, which showed a latency drift from about 15 ms initially to about 17 ms at 70 min. The short and long latencies were determined for a ganglion at 37 deg C for 35 min. This ganglion was exposed for five 1-min periods at 30 mW/sq cm (65 W/kg) alternating with 1-min control intervals during the middle 10 of the 35 min, which yielded no apparent effect on either latency. Thus, the latency drifts with time were not RFR-related.

Wachtel et al. (1975) exposed abdominal ganglia excised from the marine gastropod Aplysia to 1.5-GHz or 2.45-GHz CW RFR between center conductor and ground plane of a section of rectangular stripline, with the center conductor vertical. RFR was fed from the source through a circulator to the stripline section by coaxial cable. The section was terminated with a shorted stub, which reflected the energy not absorbed by the sample in transmission back through the sample for additional absorption. The remaining energy returned to the circulator, where it was absorbed by a dummy load. The forward and reflected powers were measured with a meter connected to a bidirectional coupler in the input coaxial cable. A few experiments were also conducted with 1.5-GHz and 2.45-GHz pulsed RFR.

Among the effects sought were RFR-induced changes in endogenous firing patterns of the two types of pacemaker neurons: "beating" pacemakers that have regular interspike intervals (ISIs) and "bursting" pacemakers, in which the action-potential (AP) bursts occur at regular interburst intervals (IBIs). Regarding pacemakers, the authors noted that the ISI of an unperturbed beating pacemaker neuron in an isolated ganglion of Aplysia usually varies less than 10% during several minutes and the IBI of a bursting pacemaker neuron was usually even more stable. Thus, RFR-induced changes exceeding about 10% were readily detected in this study, permitting determination of RFR thresholds for such changes. They also indicated that useful results were obtained from 41 separate neurons in about 25 "successful" experiments of 50.

For exposure, a ganglion was pinned with nonconductive cactus needles inside a small Plexiglas cylindrical chamber filled with seawater. The volume occupied by the various chambers used ranged from 0.3 to 0.8 ml (depending on the RFR frequency), all small relative to the volume of the stripline section. Intracellular microelectrodes were inserted through small holes in the ground-plane wall to stimulate neurons and record signals therefrom. The microelectrodes were filled with KCl solution, 2.5 M at the tip and 0.5 M in the barrel, yielding relatively low and high resistivities, respectively, with the latter rendering the barrel less susceptible to RFR and thereby reducing artifact potentials to less than 1 mV.

With the simple geometry of the stripline (which propagated transverse electromagnetic waves) and the small sample size, SARs within the sample (expressed as absorbed powers in mW/cc) were calculated from the input power to the stripline, its cross-section, the dielectric properties and size of the Plexiglas chamber, and the volume and dielectric properties of the sample, as described in the appendix to the paper. The authors estimated that the values were incorrect by no more than 20%.

Temperature rises within the sample were monitored with a thermistor probe. The authors noted that by connecting the thermistor to a bridge circuit isolated from ground, previously observed temperature jumps indicating that the thermistor was functioning as an RFR sink were eliminated.

To illustrate the RFR-threshold-determination method, firing patterns for a beating pacemaker given a steady stimulus current to set its ambient ISI to 1 per second were presented. Each pattern covered about 1 min before and 1 min after exposure to 1.5-GHz CW RFR at 15, 30, or 45 mW/cc. No change in ISI was evident for 15 mW/cc, a slight ISI increase was observed several seconds after the onset of the RFR at 30 mW/cc, and the ISI increased by about 50% at 45 mW/cc. The authors concluded that 30 mW/cc was the threshold for this effect in this neuron.

Presented also were three illustrative bursting-pacemaker patterns. The first was a control pattern (2 min) to show the regularity of the IBI. The second was a 2-min pattern during the middle of which the neuron was exposed for about 25 seconds to 1.5-GHz CW RFR at 150 mW/cc, a level selected to yield a significant temperature rise (about 1 deg C) in the preparation. With RFR onset, the temperature rose linearly, and the second pattern showed an accompanying decrease in IBI (opposite to the RFR-induced increase in ISI for beating pacemakers). On termination of the RFR, the temperature decreased more slowly, and the IBI increased correspondingly. The third pattern was obtained with convective warm-air heating for about 25 seconds to obtain a temperature rise about the same as that induced previously by the RFR. The result was opposite to that with the RFR, an IBI increase (barely perceptible in the pattern) during heating. (Not clear was whether these patterns were for three distinct bursting-pacemaker neurons or for a single neuron given the RFR and convective-heat treatments in succession.)

The ISI thresholds for 14 beating pacemaker neurons tested were about 5 mW/cc for five, 10 mW/cc for four, 15 mW/cc for one, 25 mW/cc for one, and 30 mW/cc for three. For seven bursting pacemaker neurons, the IBI thresholds were 5 and 45 mW/cc for two at each level and 10, 25, and 30 mW/cc for one at each level.

In experiments with 1.5-GHz pulsed RFR (10-microsecond pulses at 1000 to 5000 pps) at 150 mW/cc (average), the percentage increases in ISI were comparable to those for 1.5-GHz CW RFR at the same level and convective heating. The authors noted that 10-microsecond pulses were very short compared to the millisecond current pulses usually used to stimulate nerve cells, and that in a few experiments, "microwave pulses in the range of 10-100 msec may have increased efficacy in synchronizing firing patterns. A related effect has previously been reported in frog heart (Frey and Seifert, 1968)." (The latter finding is discussed in Section 3.6.3.) The authors noted that only a few experiments were performed with 2.45-GHz RFR for comparison with those with 1.5-GHz RFR and stated that: "...it is not possible to say that any difference in effects occurs at these two frequencies."

In their conclusions, the authors reiterated that the changes in firing patterns at RFR levels below 10 mW/cc were quite reproducible and they offered several inconclusive speculations regarding whether such effects would occur in cortical neurons of humans exposed at power densities between 1 and 10 mW/sq cm.

Chou and Guy (1978) designed a waveguide exposure system to study the effects of 2.45-GHz RFR on isolated tissues. Exposures were done from below in a vertical section of rectangular waveguide (inside dimensions 7.2 cm x 3.4 cm) filled with Ringer's solution to a height of 6 cm. A quarter-wavelength dielectric slab was used to match the impedance of the solution to that of air and to seal the bottom of the section. By measurement, the penetration depth of the solution was 1.65 cm at 2.45 GHz, so the 6-cm column of fluid was essentially equivalent to one of infinite length. A pump connected to inlet and outlet ports through the walls at the bottom of the section was used to circulate solution maintained externally at constant temperature.

Holes 3 mm in diameter were drilled in the four walls 1 cm above the dielectric slab, and Plexiglas chambers were glued to the outside walls against each hole. These chambers were used for holding stimulating and recording electrodes for specimens of isolated tissue mounted across the center of the waveguide either parallel or perpendicular to the electric field of the TE10 mode.

SARs of the Ringer's solution at the location of the isolated specimen were calculated from values of net forward (input minus reflected) power, the density of the solution, the cross-sectional area of the waveguide, and the attenuation constant of the solution. The authors noted that thin tissue preparations having dielectric properties close to those of the solution would not significantly perturb the fields therein. Thus, such calculated local SARs in the solution were deemed applicable to the tissues as well.

Using nonperturbing probes, SARs were also determined by measurements of temperature rises within specimens. Because of rapid heat convection in Ringer's solution, preparations were embedded in a jelly simulation of Ringer's solution. The results were in reasonable agreement with the calculated values (about 10% lower than the latter, possibly due to differences in properties of simulated and actual Ringer's solutions).

With this system, possible RFR effects on frog-sciatic and cat-saphenous nerves, which contain mainly myelinated nerves, and on the rabbit vagus nerve, which consists of both myelinated and unmyelinated fibers, were sought. Also sought were effects on the superior cervical ganglion of the rabbit because it contains not only neuron cell bodies but also numerous synaptic junctions at which acetylcholine and norepinephrine are released.

Each vagus nerve or superior cervical ganglion was mounted within the waveguide (parallel or perpendicular to the E-vector), with one end passing through an appropriate hole in the waveguide wall to a pair of stimulation electrodes and the other end through the opposing hole to recording electrodes. Before, during, and after exposure to RFR, each specimen was stimulated with a 0.3-millisecond current pulse of 0.3-30 mA at 2-second intervals. The compound action potentials (CAPs) were recorded and their conduction velocity and amplitude were determined.

Vagus nerves were exposed to 1-microsecond RFR pulses at 1000 pps or 10-microsecond pulses at 100 pps for 10-min periods at average SARs of 0.3, 3, 30, and 220 W/kg or to CW RFR at the same SARs, with 5 min between exposures. The superior cervical ganglia were exposed for only 5-min periods because of their shorter lifetimes. Specimens were also exposed to pulsed RFR at an SAR of 220 W/kg average (220 kW/kg peak) and to CW RFR at 1500 W/kg in the absence of current-pulse stimulation to test for the possibility of direct RFR stimulation.

The temperature of the solution at the fluid outlet of the waveguide section was held constant at 37 +/- 0.02 deg C during exposure. At the higher RFR levels, however, the fluid temperature at the center of the mounted specimen (measured with a nonperturbing probe) rose by as much as 1 deg C because of the limited circulation rate of the pump. Frog-sciatic and cat-saphenous nerves were studied in similar fashion, but the frog preparations were immersed in amphibian Ringer's solution held at room temperature instead of mammalian Ringer's solution at 37 deg C.

For exposures of stimulated preparations to pulsed or CW RFR in either orientation relative to the E vector at SARs that did not increase the fluid temperature near the center of the specimen, no changes in either amplitude or conduction velocity of the CAP were observed. At the SARs that increased the fluid temperature by 1 deg C, a slight increase in conduction velocity was obtained, exemplified by the conduction velocity and peak CAP amplitude vs time, shown in Fig. 5 of the paper, for a cat saphenous nerve exposed to CW RFR at 1500 W/kg parallel to the E-vector. During exposure, the temperature increased by 1 deg C and the conduction velocity increased by about 2%, but the amplitude variations displayed were small and apparently not RFR-dependent. This conduction-velocity increase was reproduced by raising the temperature of the solution 1 deg C by non-RFR means.

Figure 6 of the paper showed CAP recordings for a stimulated rabbit vagus nerve before, during, and after exposure to 10-microsecond pulses (100 pps) at 220 kW/kg peak, and to CW RFR at 1500 W/kg, with the nerve perpendicular to the E-vector. For the pulsed-RFR case, the solution temperature rose 0.3 deg C, which increased the conduction velocity slightly (from 117 to 118 m/s). For the CW case, the temperature rise of the solution was 1 deg C, which increased the conduction velocity from 117 to 135 m/s. An equivalent rise in solution temperature by non-RFR means yielded the same velocity increase.

The CAP recordings for a rabbit superior cervical ganglion mounted at right angles to the E-vector and exposed to 1-microsecond pulses (1000 pps) and to CW RFR at the same SARs as the vagus nerve were shown in Fig. 7 of the paper. For the pulsed RFR, the temperature rise of the solution was 0.3 deg C as before, but the latency time, 17 ms, remained unchanged. The temperature increase produced by the CW RFR was again 1 deg C, and the latency time decreased to 16 ms, a result reproduced by increasing the solution temperature 1 deg C by non-RFR means.

In the absence of stimulation by current pulses, no direct stimulatory effects of exposure were observed at the highest available SARs (pulsed RFR at 220 kW/kg peak or CW RFR at 1500 W/kg).

In addition to nerve preparations, Chou and Guy (1978) investigated the effects of RFR on contraction of rat diaphragm muscle. A muscle tension transducer was designed so that during contraction, a shutter between a light-emitting diode and a photoresistor would alter the transmission of infrared light from the diode to the resistor and hence the voltage across the resistor in proportion to the tension.

Diaphragm muscle with right and left phrenic nerves were excised and a section about 4 sq cm of the sternocostal portion of the muscle was dissected in Ringer's solution. Strings sutured on the central tendon and residual intercostal muscles were pulled through small opposing holes on the narrower walls of the waveguide. Two Plexiglas chambers attached to the exterior waveguide walls served to support the muscle under tension, and a third provided an access port for stimulating the phrenic nerve. The muscle was fixed at one end and connected to the tension transducer near the other end by the strings.

Current pulses 0.3 ms in width and 0.3-30 mA in amplitude were applied to the phrenic nerve once every 5 seconds for single-twitch experiments and 15-30 per second for tetanus experiments. The muscles were exposed to the pulsed or CW RFR at each SAR for 5 min, with intervals of 5-10 min between exposures.

Atransient-averaging computer was used to obtain plots of mean tension vs time for 10 single twitches. The mean single-twitch tension for a muscle before, during, and after exposure to 1-microsecond pulses (1000 pps) at 220 kW/kg peak and to CW at 1500 W/kg were displayed in Fig. 8 of the paper. The pulsed RFR increased the solution temperature 0.2 deg C, but had little effect on twitch tension. The CW RFR increased the solution temperature by 1 deg C, a rise accompanied by a decrease in tension amplitude and a reduction of latency time. The postexposure values of tension amplitude and latency time were even smaller; however, these exposure and postexposure results were replicated by a non-RFR-induced solution-temperature increase of 1 deg C.

Plots of mean tetanus tension for a muscle stimulated with 0.3-ms, 30-mA pulses at 15 pps and exposed to maximum pulsed and CW SARs (220 kW/kg and 1500 W/kg) during the decreasing phase of tetanic contraction were displayed in Fig. 9 of the paper. No RFR-induced changes were evident. Tests to determine whether the RFR at these SARs directly altered the tension of muscles not stimulated with current pulses also yielded no effects.

In their discussion, the authors noted that their negative results at SARs that did not increase the solution temperature were at variance with those of Kamenskii (1964, 1968), Rothmeier (1970), and Portela et al. (1975). However, the nerve preparations of Kamenskii and Rothmeier were exposed to RFR in air, so the SARs could have been high even though the incident power densities were low.

The authors also indicated that their negative findings in the absence of current-pulse stimulation do not support the hypothesis of direct neural stimulation proposed by Frey (1971) as the basis for the RFR-auditory effect or that RFR causes neurotransmitter release in isolated turtle hearts by excitation of nerve remnants (Tinney et al., 1976).

It is important to emphasize that the RFR fields were present within the preparations irrespective of their temperature equilibration with the Ringer's solution, i.e., even when the capabilities of the circulation system were adequate to avoid discernible temperature rises therein. Thus, a hypothesis that changes in neural CAPs or muscle tension might be produced by the fields per se (i.e., nonthermally) is negated by the absence of such effects at the SARs that yielded no measurable rises in solution temperature. On the other hand, the positive results at the SARs that increased the solution temperature were clearly thermal, as evidenced by their replication with non-RFR means. It is also likely that the preparations were heated by contact with the fluid rather than by RFR absorption, because these results were insensitive to orientation of the preparations relative to the E-vector.

The numbers of nerve and muscle preparations of each type studied were not indicated. Presumably the single results presented for each type were representative, but they provided no indication of how reproducible such results were. Representative results for the frog sciatic nerve were not presented.

McRee and Wachtel (1980) exposed isolated frog sciatic nerves to 2.45-GHz CW RFR from below in a waveguide system similar to that used by Chou and Guy (1978). Two thin-wall polyethylene tubes 2 mm in diameter were placed along a transverse centerline of the waveguide, presumably in the orientation parallel to the electric vector of the TE10 mode. One tube was located in proximity to the slab; the other was located 5 cm above the slab surface, where the intensity of the RFR was negligible because of the attenuation by the solution. One sciatic nerve from each frog was pulled through the lower tube for exposure and its mate was pulled through the other tube as control. The tubes were used to preclude effects of toxic ions in the waveguide solution and to prevent leakages that could shunt the electrodes. Both tubes were filled with Ringer's solution. Stimulation electrodes were attached at one end of each nerve outside the waveguide, and recording electrodes were attached at the opposite ends.

SARs of the Ringer's solution at the lower nerve site were calculated as in Chou and Guy (1978). To determine whether the polyethylene tubes significantly altered the absorption characteristics of the solution, temperature-vs-time profiles were measured, during exposure at 100 W/kg for 60 min, with small glass-coated thermistors at locations within the tubes, just outside the tubes, and at the tube sites in their absence. The profile for the site next to the lower tube was slightly higher than for inside that tube, and the latter profile was slightly higher than for the site in the absence of the tube, but the differences were no greater than 0.1 deg C (at 100 W/kg). The differences among profiles for the control tube were negligible.

As noted by the authors, for a short period after a nerve is stimulated to fire by a current pulse, it is relatively refractory (less responsive to a second pulse). Thus, in the absence of fatigue, presentation of two pulses spaced at an interval longer than the refractory period would yield two compound-action-potential (CAP) responses of the same shape and magnitude. However, as a nerve fatigues from repeated stimulatory activity, its refractory period increases. Therefore, the amplitudes of CAP responses to second pulses (second-CAP responses) spaced from first pulses at intervals equal to the initial refractory period diminish with time relative to the amplitudes of the CAPs for the first pulses (first-CAP responses).

Since the refractory period for a healthy frog sciatic nerve is slightly longer than 5 ms, the exposed and control nerves were both stimulated with pairs of current pulses spaced 5 ms apart, so that the second pulse would occur at almost the end of the initial refractory period. Thus, time changes of the refractory period of the control nerve could be observed concurrently with possible alterations thereof in the nerve exposed to RFR. At the beginning of each experiment, both nerves were stimulated for 10 min with pulse pairs at a repetition rate of 5 pairs per second with no RFR present, allowing the nerves to stabilize. The repetition rate was then increased to 50 pairs per second and exposure to RFR at the desired SAR was begun.

In the first series of experiments, nerves were exposed at 100, 50, and 20 W/kg for 60 min at each SAR without circulation of the solution in the waveguide. As a consequence, the temperature-time profile for the exposed nerve at each SAR was higher than for the control nerve, with a difference of about 0.5 deg C at 100 W/kg. In this series, the stimulus magnitude of the pulse pair was set to yield maximal CAP amplitude for the first pulse and half-maximal CAP amplitude for the second pulse.

Representative recordings for 50 W/kg were displayed in Fig. 6 of the paper. The amplitudes of the first and second CAPs for the exposed nerve after 5 min of exposure were slightly smaller than corresponding values for the control nerve, with minor shape differences discernible. After 43 min, the CAPs of the control nerve had diminished slightly (retaining the 2:1 ratio between first and second amplitudes) but the CAPs of the exposed nerve had diminished much more. At 55 min, further diminution of the CAPs of the control nerve had occurred but the CAPs of the exposed nerve had become barely discernible on the scale used.

To quantify CAP changes and to account for vitality changes in control nerves, the "half-decay time ratio" (HDTR) for each CAP was defined as the time necessary for the CAP of the exposed nerve to decrease to half its value, divided by the half-value time for the corresponding CAP of the control nerve. Thus, values of HDTR less than unity indicated that the exposed nerve decayed faster than the control nerve, and vice versa for HDTRs larger than unity.

The HDTRs for the first series were given in Table I of the paper (adapted and presented below as Table 23).


5-23 100 0.99 0.48
5-24 100 0.56 0.35
8-30 100 0.80 0.83
9-8 100 1.00 0.25
Means 0.79 0.39
6-1 50 0.96 0.79
6-7 50 0.82 0.50
6-8 50 0.79 0.86
6-17 50 0.23 0.21
9-18 50 0.71 0.78
9-19 50 0.40 0.61
Means 0.51 0,49
6-14 20 1.00 0.59
6-16 20 0.67 0.50
9-20 20 1.00 1.00
9-22 20 0.70 0.74
Means 0.82 0.66


The authors stated: "The times for the CAPs of the exposed nerves to decay to half amplitude for all SARs ... were less than for the CAPs of the unexposed nerves. Although no clear dose-response relationship could be detected for the first CAP decay, a faster decrease in amplitude of the second CAP did occur with increasing dose (SAR)."

It is noteworthy that presentation of the HDTRs for each nerve in Table 23 permitted not only a check of the means but also calculation of the standard deviations (SDs), not given in the paper. Surprisingly, all of the means shown in Table 23 were in error. The correct means and SDs are shown in Table 24:


100 0.84 +/- 0.21 0.48 +/- 0.25
50 0.65 +/- 0.28 0.63 +/- 0.24
20 0.84 +/- 0.18 0.71 +/- 0.22


The considerable variations among the individual results for each SAR were probably not RFR-related. By using the corrected values, dose-dependency was examined with the 1-tailed t-test, which showed that the differences among the first-CAP means for 100, 50, and 20 W/kg were statistically nonsignificant (p>0.05). This was also true for the second CAPs of the first series. At 100 W/kg, however, the mean second-CAP HDTR was significantly lower than the mean first-CAP HDTR.

Also performed in the first series were three experiments in which the temperature gradient between exposed and control nerves was reversed by using an infrared (IR) source to heat the control nerve (in the absence of RFR) to obtain the same differential level as that with RFR (but with the top nerve hotter than the bottom one). A reasonable simulation of the 100-W/kg temperature profile was achieved for the upper nerve in this manner. No CAP results for the IR experiments were presented, but the authors indicated that there was no significant difference in CAP decay time between the upper (IR-heated) and the lower (unheated) nerves. They therefore concluded: "From these results it would appear that the effects obtained are not due to a difference in temperature of the nerves but are specific to the microwave radiation." In their discussion, however, they stated: "Although elevating the temperature of the nerve did not have the same effect on vitality as microwaves, the conclusion that the effect on vitality is microwave specific does not preclude the possibility that nonuniform, localized heating or thermal gradients inside the nerves are the mechanisms producing the effect."

In the second series of experiments, the Ringer's solution within the waveguide was circulated through a temperature-controlled water bath. In this manner, the exposed and control nerves were maintained at 24 +/- 0.05 deg C for all SARs. The stimulus was set to produce equal maximal first and second CAPs. Exposures were done at 0, 5, 10, and 20 W/kg but only the results for 0, 5, and 10 W/kg were presented. (Two experiments of the second series were done at 20 W/kg; no data were presented, but the authors stated that since the results were basically the same as those of the first series at 20 W/kg, higher SARs were not used.)

Typical recordings for exposure 2, 50, 105, and 172 min at 10 W/kg were presented in Fig. 7 of the paper. At 2 min, there was little difference between the first and second CAPs for either nerve or between the two nerves. At 50 min, the second CAP of the control nerve had diminished relative to its first CAP, indicating that the control nerve had lost some vitality. However, this effect was more pronounced for the exposed nerve. By 105 min, the first CAPs of both nerves and the second CAP of the control nerve had diminished, but the second CAP of the exposed nerve was no longer evident. By 172 min, the first and second CAPs of the control nerve had both decreased considerably, but both CAPs of the exposed nerve were absent.

The HDTRs obtained in the second series are shown in Table 25 (adapted from Table II of the paper). A check showed the means to be correct; the SDs are also shown.


5-24 0 1.40 1.57
7-11 0 1.21 1.00
7-13 0 1.11 1.04
Means and SD's 1.24 +/- 0.15 1.20 +/- 0.32
6-28 5 1.21 1.19
7-05 5 0.75 0.68
7-07 5 1.91 0.97
7-16 5 1.07 1.22
Means and SD's 1.24 +/- 0.49 1.02 +/- 0.25
5-19 10 0.64 0.50
6-21 10 0.35 0.55
6-22 10 0.36 0.40
7-12 10 0.74 0.17
Means and SD's 0.52 +/- 0.20 0.41 +/- 0.17


In their discussion, the authors concluded that the SAR threshold was between 5 and 10 W/kg and that the effect was not reversible, since on termination of exposure, the nerves did not revitalize or increase their activity above that at the end of exposure.

It is interesting that the first-CAP means for 0 and 5 W/kg were both much larger than unity (1.24), indicating that the control nerves had decayed faster than the exposed nerves; however, the mean value at 10 W/kg was less than unity (0.52) and the decrease was significant by the 1-tailed t-test. Also, the second-CAP mean was larger than unity at 0 W/kg, about unity at 5 W/kg, and less than unity at 10 W/kg; the means for 0 and 5 W/kg did not differ significantly, but the mean for 10 W/kg was significantly lower than for 5 W/kg. Thus, these results support the existence of a threshold between 5 and 10 W/kg, but the occurrence of HDTRs larger than unity indicates that uncontrolled non-RFR factors were present (ascribed by the authors to "natural variability of living nerves"). On the other hand, none of the HDTRs in the corrected results for the first series (Table 24) exceeded unity and all of the means were much less than unity, but the means showed no statistically significant SAR dependence (20 W/kg and higher). Thus, it is difficult to determine the relative importance of RFR and non-RFR factors in this study.

In a subsequent study, McRee and Wachtel (1982) used 2.45-GHz pulsed RFR (10-microsecond pulses at 50 pps) at an average SAR of 10 W/kg (20 kW/kg peak) in the same exposure system to study and compare the effects on frog sciatic nerves with those obtained previously with 2.45-GHz CW RFR at 10 W/kg (McRee and Wachtel, 1980).

In this as in the previous study, the stimulus was set such that the initial first and second CAPs were maximal and approximately equal in magnitude. At the start of each experiment, both nerves were stimulated for 10 min in the absence of RFR with pulse pairs at a repetition rate of 5 pairs per second, which allowed the nerves to stabilize. The repetition rate was then increased to 50 pairs per second (i.e., twin-pulse stimulation was done at 20-ms intervals) and RFR exposure was begun (at 10 W/kg average).

The 10-microsecond RFR pulses were delivered at 50 pulses per second, i.e., at 20-ms intervals. In one set of experiments, the RFR pulses were synchronous with the peak of the CAP. In another set, they were timed to arrive during the 15-ms quiescent intervals between successive 5-ms twin-pulse stimulations. In still another set, the RFR pulses were made to arrive asynchronously, i.e., at various times during the 5-ms stimulation intervals.

The authors did not present any representative CAP recordings, but they described the results qualitatively as follows: "During exposure the exposed nerves first underwent a prolongation of their refractory period as evidenced by a decrease in amplitude of the second CAP. This prolongation of the refractory period and decrease in the second CAP usually were observable after 20 to 30 min of exposure. Later in the exposure, severe decreases in maximal CAP amplitude occurred. During the same periods the control nerves showed little change in these characteristics."

As in the previous study, the HDTR (half-decay time ratio) was used to quantify RFR-induced effects on CAPs and to account for vitality changes in control nerves. The HDTRs for the first and second CAPs derived from five experiments with RFR pulses delivered in phase with the peak of the CAP are shown in Table 26 (Table I of this paper, in which the authors did include the SDs).



12-3 0.72 0.74
12-7 0.76 0.71
12-12 0.35 0.60
12-14 0.26 0.43
12-19 0.30 0.49
Means and SD's 0.58 +/- 0.25 0.59 +/- 0.14


Surprisingly again, there were errors: the correct first-CAP mean and SD are 0.48 +/- 0.24; the second-CAP mean was correct but the correct SD is 0.13. However, these errors were not material because the difference in means was not significant in either case. All 10 individual ratios were less than unity, indicating that both the first and second CAP of each exposed nerve had decayed faster than its control nerve.

Tables 27 and 28 (Tables II and III of the paper) displayed the mean first- and second-CAP ratios and SDs for the five experiments with synchronous RFR pulses out of phase with the peak of the CAP and for the six experiments with asynchronous RFR pulses, respectively.



12-31 0.85 0.71
1-3 0.55 0.57
1-4A 0.93 0.73
1-4B 0.53 0.60
2-25 0.46 0.63
Means and SD's 0.66 +/- 0.21 0.69 +/- 0.07



2-26A 0.82 0.83
2-26B 0.31 0.93
2-27A 0.60 0.71
2-27B 0.36 0.33
2-28 0.44 0.61
2-29 0.45 0.76
Means and SD's 0.50 +/- 0.21 0.65 +/- 0.21


The means and SDs in Table 27 were correct and small errors in Table 28 were inconsequential. In both sets of experiments, all the ratios were also less than unity, but the first-CAP and second-CAP means did not differ significantly. The authors stated: "In all cases both CAPs of the exposed nerves lost their vitality in a shorter time than in the control nerves. Statistical examination of the data was performed using both the analysis of variance and the two-sided t-test. In all cases the loss of vitality of the exposed nerve was highly significant (P<=0.01) when compared to that of the control nerve. However, no significant difference in loss of vitality was detected with different phasing of the microwave pulses with the CAPs." A minor comment about the statistical treatment is that perhaps the authors should have used the 1-tailed instead of the 2-tailed t-test, because the outcome sought was unidirectional (loss of nerve vitality).

The less-than-unity HDTRs displayed in the three tables of the paper clearly indicate that the vitality of every exposed nerve, as assessed from the first and second CAPs separately, diminished more rapidly than that of its control nerve. Not clear, however, is the interpretation of the observation that some of the second-CAP HDTRs were larger than their corresponding first-CAP HDTRs. For example, the first-CAP and second-CAP HDTRs obtained in one of the experiments with RFR pulses at the peak of the CAP response were respectively 0.35 and 0.60 (Table 26). One possible interpretation of this result is that the second-CAP time of the control nerve had not changed materially relative to its first-CAP time (as befits an adequate control) and that the second-CAP time of the exposed nerve had increased about 71% relative to its first-CAP time, a biologically unlikely result. A more tenable interpretation is that the second-CAP time of the control nerve had decreased materially while that of the exposed nerve also had decreased, but less so. If this had been the case, then there were uncontrolled non-RFR factors present in the experiment.

The case above is not unique. Specifically, the second-CAP mean HDTR for that set of experiments (0.59) was 23% larger than the (corrected) first-CAP mean (0.48); also, the first- and second-CAP means shown in Table 28 were 0.50 and 0.69, respectively. It is interesting that in the study with CW RFR (McRee and Wachtel, 1980), there were also a few isolated cases similar to those above, but all of the second-CAP means were smaller than their corresponding first-CAP means.

On the overall findings of this study, the authors noted: "It would seem reasonable that if pulse-microwave radiation had an immediate effect on the sciatic nerve, it would have been greater with the pulses delivered during the firing of the nerve. Our results showed that the increased rate of loss of vitality of the exposed nerves did not depend upon the phasing of the microwave pulses with the nerve action potential. Asynchronous microwave pulses which moved randomly through the action potential, pulses synchronized with the peak of the action potential, or pulses synchronized with the quiescent period of the action potential produced the same effect on vitality." They also noted that by analysis of variance, there were no significant differences in the effects on nerve vitality of 2.45-GHz CW and pulsed RFR at 10 W/kg.

The findings of the two studies by McRee and Wachtel (1980,1982) appear to be contrary to the negative results reported by Chou and Guy (1978) for exposure of isolated nerves from the cat and rabbit to CW and pulsed 2.45-GHz RFR. As previously noted, however, Chou and Guy (1978) did not present any data for the frog. IN-VITRO HISTOCHEMICAL EFFECTS

Olcerst and Rabinowitz (1978) performed experiments to determine whether 2.45-GHz RFR alters the activity of the enzyme acetylcholinesterase (AChE) in purified form (at a concentration of 5 units per ml in 0.1 M phosphate buffer) or the activity of AChE in defibrinated rabbit blood. Enzyme assays of purified samples at room temperature and blood samples at 37 deg C were performed by recording spectrophotometer. The authors noted that changes in enzyme activity could affect the concentration of bivalent cations in blood serum. Such assays were performed by atomic absorption spectroscopy.

In each experiment, two 1.5-ml samples, each within a jacketed quartz cell, were placed within an anechoic chamber, one sample for exposure at a distance of 12.7 cm (about one wavelength) from the crossover of a horizontally directed diathermy antenna and the other sample at a site that was shielded from the RFR as a control. The exposure levels were calibrated by measuring the power densities at various distances from the crossover with an NBS power density meter for several values of source power (determined with a bidirectional coupler). For cooling during exposure, paraffin oil (dielectric constant about 2.0) was pumped in series through the two jackets to an external water bath maintained at 37 deg C. Sample temperatures were not monitored during exposure but were determined from cooling curves obtained after exposure termination with a thermistor probe. The authors indicated that the temperatures of control samples tracked those of the exposed samples to 0.5 deg C. In some experiments, samples were not cooled. Control runs with the RFR off were also made.

In one set of experiments, the mean initial velocity of enzyme activity (the amount of substrate cleaved per unit time just after exposure) was determined for samples of purified enzyme in buffer exposed to 2.45-GHz CW RFR at 0, 10, 15, 25, 50, 75, 100, or 125 mW/sq cm for 0.5 hr, or at 0 or 25 mW/sq cm for 3 hr. The results were tabulated as means and SDs and whether or not each result was statistically significant at the 95% confidence level (but the numbers of samples tested were not presented). Exposure at 125 mW/sq cm for 0.5 hr with the sample uncooled yielded the only result stated to be significant: a drop in mean initial velocity of enzyme activity.

In another set of experiments, the initial velocity of enzyme activity in defibrinated rabbit blood (the amount of substrate cleaved initially per unit time per red blood cell counted with a hemocytometer) was determined for samples maintained at 37 +/- 0.5 deg C and exposed for 3 hr to 2.45-GHz CW or square-wave-modulated RFR (0.75-ms pulses at 710 pps) at 0, 21, 35, and 64 mW/sq cm (average power densities). Tabulated were results that appeared to be for one sample per exposure condition, i.e., with no indication of statistical treatment. These results showed no differences.

The release of bound calcium and magnesium from rabbit blood cells was investigated in a third set of experiments. Samples were sham-exposed or exposed to 2.45-GHz RFR (presumably CW) for 3 hr at 25 mW/sq cm, after which the serum concentrations of Ca and Mg ions were assayed. The means and SDs for each cation were presented with an indication that by t-test, none of the differences between exposed and control samples was significant (but the numbers of samples tested were not given).

Regarding their only positive finding, the authors noted that with the samples uncooled, the RFR power absorbed was sufficient to denature the enzyme, and that therefore the result was not relevant to possible in-vivo effects on AChE. However, they indicated their awareness of the limitations of postexposure assays.

Galvin et al. (1981c) investigated whether the activities of AChE and creatine phosphokinase (CPK) are altered by in-vitro exposure to RFR. They exposed AChE derived from the electric eel or CPK derived from rabbit muscle to 2.45-GHz CW RFR in a chamber consisting of a distilled-water-filled waveguide section. The input end of the section was sealed with a dielectric slab of quarter-wavelength thickness, to match the impedance of the section to that of air, and the output end was sealed with a short. For exposure to the RFR, a cylindrical tube holding 4.2 ml of sample was inserted adjacent to the dielectric slab through holes in the waveguide section (with the tube ends outside the waveguide); for control, another tube of sample was similarly inserted 9.5 cm along the propagation direction from the first tube, at which the RFR level was negligible because of the attenuation by the intervening water. The contents of each tube were stirred magnetically outside the waveguide and their temperatures were maintained at 37.25 +/- 0.25 deg C by water circulating continuously through the waveguide.

For treatment of AChE, 0.025 ml of the enzyme was added to 4.0 ml of phosphate buffer within each tube and the mixture was incubated for 5 min prior to exposure to equilibrate to 37 deg C within the waveguide. For treatment of CPK, 4.0 ml of substrate solution (CPK -1) in each tube was allowed to equilibrate for 5 min, 0.2 ml of CPK was then added, and the RFR was turned on. Exposures of each enzyme were at SARs of 1, 10, 50, or 100 W/kg. At the start of exposure and at every 2 min afterward for 10 min, 0.7-ml aliquots were taken from each tube and placed in test tubes at 0 deg C to stop the reaction. To maintain uniform exposure of the enzyme, each aliquot removed from the tubes was replaced with 0.7 ml of phosphate buffer.

Absorbance values for the aliquots of AChE and CPK were determined by spectrophotometry at 412 and 340 nm, respectively, with corrections for progressive dilution by buffer replacement of the withdrawn aliquots. Least-squares regression analysis was used on the data to obtain best-fit lines, and the paired 2-tailed t-test was used to compare reaction rates of each RFR-exposed enzyme with its control. In an auxiliary experiment, the AChE and CPK activities of unexposed diluted samples (corrected for dilution) were compared with those of unexposed undiluted samples; the differences were found to be nonsignificant (p>0.05), thus confirming the validity of successive dilution in exposure experiments.

Typical plots of absorbance vs time for samples exposed at 50 W/kg and corresponding control samples showed almost coincident linear rises for each enzyme. The AChE and CPK activities for samples exposed at 1, 10, 50, and 100 W/kg, expressed as time rate of change in absorbance, did not differ significantly from their respective controls or from each other, thus revealing no effect of the RFR at any of the levels used.

Millar et al. (1984) sought possible changes in AChE from exposure to discrete RFR frequencies in the range 2.375-2.700 GHz, using a system designed to permit optical measurements of enzyme activity in a sample during exposure. The exposure system consisted basically of a microwave stripline with its short axis vertical. Mounted thereon was an 8-mm-square sample cell 3 mm thick, with one diagonal vertical and pipes fitted thereto to pump sample fluid into and out of the cell and to permit escape of bubbles in the sample. The normal flow rate was 10 ml/min, the sample volume was 0.192 ml, and the total volume of mixture circulated was about 4 ml, so the normal volume element residence time (cell volume/flow rate) was 0.0192 min (not seconds, as stated at the top of p. 171) or 1.15 seconds. SARs were calculated from measurements of forward, reflected, and transmitted powers and the sample volume.

For measurements of enzyme activity during exposure, a light beam from a monochromator (450 nm) was piped to the center of one of the sides of the cell by a quartz fiber-optic bundle. Light emerging at 180 deg from the center of the opposite side of the cell was similarly conveyed to another monochromator (450 nm) and thence to a photomultiplier with output fed to a photometer. Light emerging from the center of one of the other sides of the cell, scattered at 90 deg to the incoming beam by the sample, was similarly treated. Enzyme activities were determined alternately in the presence and absence of RFR with four fresh samples each, and the results were expressed as mean ratios (MRs) of activity with RFR present to activity with RFR absent and the SDs.

The AChE was derived from the electroplax of the ray fish, Narcine brasiliensis, and was rendered membrane-free. For most experiments, 50 microliters of enzyme mixed with about 20 ml of reaction solution was recirculated through the cell and through temperature-controlled jackets at 10 ml/min, which held the mixture at 25 or 37 deg C (+/- 0.2 deg C).

Exposures in one series of experiments were to 2.45-GHz pulsed RFR, with pulse durations of 4.8, 2.4, and 1.2 microseconds at repetition rates of 625, 1250, and 2500 pps, respectively, to yield a duty cycle of 0.003. About the mean (time-averaged) SARs for this series, the authors stated: "...the mean specific absorption rate (SAR) at the two absorbed power levels can be calculated as 2,460 W/kg and 4,290 W/kg, respectively. Peak power SAR is then 0.82 MW/kg and 1.43 MW/kg." The latter values are consistent with the stated duty cycle. In Table 29 (adapted from Table 1 of the paper), which embodied the results for this series, the mean SARs cited were 2,460 W/kg and 4 (not 4,290) W/kg, a discrepancy that could not be resolved because the authors did not state the two absorbed power levels alluded to above.


Mean SAR = 2460 W/kg Mean SAR= 4 W/kg
Pulse Width 4.8 2.4 1.2 4.8 2.4 1.2
Rep. Rate 625 1250 2500 625 1250 2500
MR at 25 deg C 1.00 1.00 1.03 0.98 1.04 1.01
+/- SD 0.012 0.026 0.018 0.036 0.003 0.010
MR at 37 deg C 1.00 0.980 0.985 1.00 0.933 1.00
+/- SD 0.001 0.020 0.010 0.005 0.012 0.003


All of the MRs were essentially unity, and the authors concluded: "At these levels, there is no significant effect of microwave irradiation on AChE activity."

In the next series of experiments, the effects on AChE activity of 2.45-GHz RFR at relatively low SARs were determined. Pulses 8 microseconds in duration at 50 pps were used (0.0004 duty cycle), and the temperature of the samples was 25 deg C. The results are displayed in Table 30, in which the mean absorbed powers (in mW) as well as SARs are stated:



Mean Absorbed Power 68 36 6.9 2.6
Mean SAR 330 180 34 13
MR 0.972 0.967 1.005 1.00
+/- SD 0.007 0.002 0.016 0.003

The authors indicated that by paired t-test, only the results for 330 and 180 W/kg (a loss of about 3% in AChE activity) were significantly different from unity (p<0.001). However, they questioned whether such a loss would measurably affect behavior or neurologic function.

Within the limits of the available RFR generator, possible effects of RFR frequency were sought. Pulses 16.7 microseconds in duration at 30 pps were used (0.0005 duty cycle), and the temperature of the samples was 25 deg C. The results, presented in Table 31, showed no significant departures from unity:



Frequency (GHz) 2.375 2.425 2.500 2.550 2.625 2.700
Mean SAR 42 40 42 44 35 32
MR 0.999 0.999 0.998 0.997 1.002 0.988
+/- SD 0.0003 0.003 0.002 0.002 0.004 0.003

Frequency (GHz) 2.375 2.425 2.500 2.550 2.625 2.700

Mean SAR 42 40 42 44 35 32

MR 0.999 0.999 0.998 0.997 1.002 0.998

+/- SD 0.003 0.003 0.002 0.002 0.004 0.003


Next, repetition rates of 10-90 pps (in steps of 10 pps) at a constant duty cycle of 0.0004 were used to determine whether this variable was important. The mean SARs were all about 1900 W/kg and the sample temperature was 25 deg C. The MRs were presented graphically with SD bars with a view toward more readily revealing "EMR-effect windows," if present. [See Section 3.4.4.] From the graph, the minimum and maximum MRs were about 0.97 (at 50 pps) and 1.00 (at 40 pps) and most of the SD bars overlapped.

The authors stated that there was no distinguishable trend of increase or decrease in AChE activity, but since all but one MR was less than unity, the sign test indicated that: "the apparent treatment-associated decrease is significant within 95% confidence limits." They also noted, however: "While an oscillatory mode might be concealed in these data, the fact that the standard error bars overlap make us question the sign test as validly pointing to a significant EMR effect on enzyme activity. This view was supported by a Student's paired t-test that showed that only the 50- to 60-pulse experiments achieved confidence levels of .05 and .01, respectively. The remaining values clustered about confidence levels of .2--not significant. Nevertheless, the 50- to 60-pulse/s results are statistically significant. They fall in the exposure rates of known EMR windows, but the biological consequences of such small changes are moot at best."

The possibility that duty cycle might be important was considered also. For 2.45-GHz RFR at 1750 W/kg with pulses at 50 pps and duty cycles of 0.0008, 0.0016, and 0.0032, no change in AChE activity was observed; the MR was 0.996 +/- 0.005.

Last, the authors suggested the possibility (deemed unlikely) of RFR-initiated kinetic-reaction chains, particularly with regard to residence time of samples within the exposure cell, which might lead to cumulative or time-delayed effects. Experiments were done with 1.2-microsecond pulses at 2500 pps for a mean SAR of 483 W/kg, in which only the flow rate of mixture through the cell was varied. For residence times (cell volume/flow rate) of 0.44 to 2.09 seconds, MRs ranged nonmonotonically from 0.996 +/- 0.004 to 0.999 +/- 0.003, and the MR for zero flow rate was 1.00 +/- 0.01. IN-VIVO HISTOLOGICAL AND HISTOCHEMICAL STUDIES

Tolgskaya and Gordon (1973) reported various effects in approximately 650 animals, predominantly rats, of exposure to RFR in the range 500 kHz to 100 GHz. The pathological effects of RFR in the "decimeter band" (0.5-1 GHz) at "high intensity" (20 to 240 mW/sq cm) included multiple perivascular hemorrhages in the brain and other organs, apical-dendrite degeneration in the cortex, cloudy swelling of cytoplasm, cytoplasmic shrinkage, vacuole formation, unevenness of staining, disappearance of cytoplasmic structures, fatty degeneration, ribonucleoprotein decrease, and occasional karyocytolysis. These high intensities were capable of causing death of the animals since they produced signs of hyperthermia (temperature increases up to 42-45 deg C) in several minutes to several hours. Photographs of the exposure arrangement showed multiple animal exposures at the same time in a room that appeared to not have RFR-absorbing material on the walls. It is likely that the SARs for the individual animals under these conditions varied widely and that all effects were thermal in nature.

Exposures referred to as "low-intensity" were also performed. The authors defined the threshold field intensity for nonthermal effects ("intensity not raising body temperature") for decimeter microwaves as 40 mW/sq cm, and these exposures were generally at or slightly below 10 mW/sq cm for 60 min daily for 10 months. Investigation of the animals by ordinary morphological methods revealed practically no vascular disorders in the nervous system. "Delicate elective neurohistological methods" (not described), however, showed disappearance of spines from cortical dendrites; appearance of beading and irregular thickening of dendrites; swelling of cytoplasm of individual cells (with appearance of vacuoles) in the basal ganglia and hypothalamus; and focal and diffuse proliferation of microglial cells, with microglial processes showing initial signs of degeneration.

Many of these low-intensity effects were similar to those described for the high-intensity exposures. At 10 mW/sq cm, the whole-body SAR for a medium rat in the decimeter band (0.5-1 GHz) ranges from 6 to 8 W/kg (Durney et al., 1978, p. 95), and there were internal regions of higher local SARs that may have been enhanced in the multiple-animal exposure arrangement mentioned above. Thus, it seems likely that the described effects (more subtle than those of frank hyperthermia) were thermally induced also.

Shtverak et al. (1974) investigated whether 2.85-GHz pulsed RFR (2.7-microsecond pulses at 357 pps) at 30 mW/sq cm (average) would affect the responses to stimulation by bell of rats with an inherited disposition to sound-induced epileptic seizures. Starting on postnatal day 2, 24 such rats were exposed concurrently in perforated Perspex boxes to the RFR 4 hr daily (0800-1200) for 10 weeks excluding Saturdays and Sundays, and 16 were sham-exposed. (The exposure arrangement and animal spacing were not described, but the authors stated that the positions of the rats were changed stepwise for successive periods to obtain comparable exposures.) The absorption rates were not given, but the corresponding whole-body SAR of a small rat in either the E- or H-orientation would be about 9 W/kg (Durney et al., p. 94). During exposure, temperatures were monitored in two fixed boxes and were found to vary from 24.7 to 27.8 deg C and the temperature in the exposure room rose from 23 to 26 deg C.

Starting with postnatal week 4, the rats were weighed weekly and were tested with a bell once weekly for six weeks. If no change in behavior was observed after 1 min of uninterrupted ringing, the reaction of the rat was scored as null. For rats that responded with a motor reaction that ended in convulsions, the interval between initiation of ringing and onset of seizure was recorded. (Neither the duration of seizure nor the duration for recovery was assessed.) In addition, the number of rats that died was recorded. The results were expressed as percentages and were evaluated by t-test.

From postnatal weeks 4 to 10, the body weight of all the rats tended to increase, more so by males than females. Most mean weekly weights of the RFR-exposed males were smaller than for the sham-exposed males and similarly for the females, but the differences were nonsignificant. Two RFR-exposed rats and one sham-exposed rat died during the first 4 weeks.

Weekly sound tests yielded null reactions from about 70% of the RFR-exposed and about 20% of the sham-exposed rats, a significant difference (p<0.05). For those that responded positively, the mean time interval to seizure onset was higher in the RFR-exposed than sham-exposed rats. The authors stated that the difference between RFR and sham groups for week 6 was significant and not significant for the other weeks, but the error bars on the graph appear to overlap considerably for all weeks.

The authors concluded that exposure to the pulsed RFR tended to lower the sensitivity of rats to audiogenic seizure and increase the number of null reactions. They also noted that qualitatively similar results were obtained for rats exposed to a pulsed electrostatic field (800-V 10-microsecond pulses, 769.2 pps, at 130 V/cm, thus presumably exposed between electrodes spaced about 6 cm apart).

Albert and DeSantis (1975) sham-exposed or exposed a total of 60 adult male and female Chinese hamsters in individual vented Plexiglas chambers to far-field 2.45-GHz CW RFR within an anechoic chamber at 50 mW/sq cm for 30 min to 24 hr, or at 25 mW/sq cm 14 hr/day for 22 days. Hamsters exposed at 25 mW/sq cm were given one pellet of chow and 5 ml of water during each 14-hr exposure and ample food and water between exposures. Right after exposure at either level or after recovery for 1 to 2 weeks postexposure, hamsters were anesthetized and perfused with fixatives, and selected tissues were sectioned and stained for light- or electron-microscopic examination. Two of the animals were fixed with potassium dichromate and the brains were stained for showing degenerating myelin. Frozen sections of others were stained for revealing degenerating axons.

At the light-microscopic level, the cytoplasm of hypothalamic neurons from hamsters exposed at 50 mW/sq cm was markedly pale and displayed vacuolization and chromatolysis. Although the nuclear structure was unaltered, the cytoplasm often appeared frothy (probably an artifact called "frothy bloat," not seen by Albert in subsequent studies with modified techniques). In rare instances, vacuoles were detected within nuclei. Vacuoles were not present in the cytoplasm or nuclei of glial cells. At the electron-microscopic level, the preferential appearance of vacuoles in the cytoplasm was evident for hamsters exposed at 25 mW/sq cm for 22 days. Vacuolization and chromatolysis in hypothalamic neurons were more prominent in animals euthanized immediately after exposure than in those euthanized 6-10 days after exposure, indicating some recovery.

Similar morphologic changes were observed in the subthalamus, but not in areas adjacent to the thalamus, i.e., the ventrobasal complex and the lateral geniculate nucleus. Also, the neurons in other regions of the central nervous system of hamsters exposed to either RFR level appeared unaltered relative to comparable tissues from control hamsters. At the light-microscopic level, no changes were found in Purkinje-cell bodies of the cerebellum, motor neurons in the ventral horn of the spinal cord, mesencephalic nuclei of the trigeminal nerve, or hippocampal pyramidal neurons. (Observations of these areas by electron microscopy were not reported in this paper.)

The authors reported some tentative evidence for axonal degeneration in RFR-exposed hamsters, but indicated that vagaries may occur with the staining method used; they stated: "The degenerated axons were widely scattered and did not seem to be confined to well-delineated pathways or distinct nuclear groups, as is customarily the case in animals with known lesions."

There was no evidence of myelin degeneration, gliosis, hemorrhage, or perivascular edema in the RFR-exposed hamsters. However, vascular stasis, or congestion, was observed in some animals.

The authors did not discuss SARs, but in the discussion printed with the paper, Dr. A.W. Guy noted that SARs as high as 4 W/kg per incident mW/sq cm were measured in his laboratory for animals of similar size, that the peak SARs (at 10 to 50 mW/sq cm) could have been 40 to 200 W/kg in some brain regions of the animals, and that this SAR range far exceeds that normally used for diathermy treatment in 20-min exposures of patients. He also noted that rectal temperature would not necessarily indicate the presence of high SARs in such local areas.

Dr. J.W. Frazer queried: "How do your histologic findings compare to those seen in animals that have experienced fevers of 4-5 deg C?" Dr. Albert's response was that: "Vacuolation and chromatolysis in the hypothalamus are typical responses of heated neurons."

Albert and DeSantis (1976) studied histological effects in the brains of 30 Chinese hamsters exposed to 1.7-GHz CW RFR at 10 or 25 mW/sq cm for 30 to 120 min. For each RFR level, each of 15 hamsters was paired with a sham-exposed hamster and both were otherwise similarly treated. Of the 15 hamsters in each RFR and sham group, 11 were euthanized right after exposure and the remaining 4 of each group were euthanized after 13-15 days for recovery. The whole animal was fixed and its brain was removed, further fixed, transversely cut into four blocks that included, respectively: forebrain, diencephalon and midbrain, pons and cerebellum, and medulla. Each block was dehydrated, embedded in paraffin, sectioned and stained with hematoxylin-eosin or thionin. Serial sections were made through the diencephalon. Corresponding sections from each of the paired exposed and control hamsters were placed on the same slide prior to staining.

Eight of the hamsters exposed at 10 mW/sq cm and euthanized immediately after exposure exhibited greater cytological alterations in hypothalamic and subthalamic neurons than controls; for two of the other three, the appearance of these regions was similar to that of the controls; in the third exposed hamster, the histological changes were inconsistent. Of the four hamsters exposed at 10 mW/sq cm and euthanized after recovery, three displayed histological changes in the hypothalamus and subthalamus similar to those seen in the exposed hamsters euthanized right after exposure, and the fourth showed no changes relative to its control.

Seven of the 11 hamsters exposed at 25 mW/sq cm and euthanized right after exposure showed more extensive changes in the hypothalamic and subthalamic neurons than their controls; one hamster exhibited no differences and the other three an inconsistent histological pattern. Of the four hamsters allowed to recover after exposure, three showed more prominent changes than their controls and the other was similar to its control.

Regarding the hamsters exposed at 10 mW/sq cm, the authors noted: "Neurons of the hypothalamus and subthalamus were most dramatically affected in experimental animals that showed structural changes. More specifically, neurons of the medial part of the ventromedial as well as the dorsomedial and often the lateral nuclei of the hypothalamus were affected by irradiation. Areas of the subthalamus that were involved included the zona incerta and tegmental field, but not the subthalamic nucleus itself. Neurons of animals exposed to 10 mW/sq cm were swollen, their cytoplasm appeared vacuolated and stained less basophilic than their control counterparts...In all cases the cytological changes appeared to be restricted to the cytoplasm. Nuclei of neurons appeared normal. No evidence of pyknosis or eccentric position on the part of nuclei was observed. Similar morphological observations were made on brains of animals exposed to 25 mW/sq cm."

The authors indicated that the blood vessels in the affected brain areas appeared normal, with no evidence of hemorrhage or edema, and that no obvious histological alterations were seen in other parts of the central nervous system examined, including the cerebral cortex, cerebellum, medulla, and pons. They also noted that the findings were similar to the previous ones for 2.45-GHz RFR at 25 and 50 mW/sq cm (Albert and DeSantis, 1975), and stated: "Clearly, at those power densities one could suspect that the common factor might be the thermal effects. However, similar changes have now been observed at 10 mW/sq cm, and this power density may be nonthermal." The latter comment is questionable in the light of Guy's point above.

Albert et al. (1981a) sought effects of prenatal and postnatal exposure of Sprague-Dawley rats to RFR on the Purkinje cells of the cerebellum. The authors noted that development of the mammalian brain is dependent on migration of nerve cells, the patterns of which are well understood for the cerebellum. Cerebellar Purkinje cells were selected for study because these cells provide a relay station for major input to, and output from, the cerebellum and because they are readily identifiable in the interface between the molecular and granular layers.

In experiment 1, six pregnant rats in individual ventilated Plexiglas containers were exposed concurrently in groups of three from above with a truncated horn to 2.45-GHz CW RFR at 10 mW/sq cm in one of two anechoic chambers. Six other rats were similarly treated but sham-exposed in the other anechoic chamber. Exposures were for 5 days, 21 hr/day, starting on gestation day 17. The rats were moved for 1.5 hr twice daily during each exposure period to conventional rat cages, where food and water were available ad libitum.

In experiment 2, six matched pairs of 6-day-old litter mates were used. One of each pair in a ventilated Plexiglas cage was exposed to 2.45-GHz CW RFR for 5 days, 7 hr/day, in one of the anechoic chambers while the other was sham-exposed in the other chamber. Daily exposure sessions were for 3.5 hr each in the morning and afternoon, with 1.5 hr between sessions for feeding. During all periods of no RFR- or sham-exposure, the pups were reunited with their dams in conventional cages.

In experiments 1 and 2, the rats were transported from George Washington University (GWU) to the Bureau of Radiological Health (BRH) for exposure and were returned to GWU for study after completion thereof. For these experiments, a nonperturbing probe was used to measure power densities within each empty Plexiglas cage with and without subjects in the other cages. The results indicated that the values varied with site within the cage and with time, due to rat movements in the other cages (no data presented). Thus, it was determined that at a mean power density of 10 mW/sq cm, the power density might vary with time from 4 to 30 mW/sq cm. From Durney et al. (1978), pp. 95-96, the corresponding time-averaged SAR for each pregnant rat in either the E- or H-orientation was about 2 W/kg, with a range from 0.8 to 6 W/kg.

In experiment 3, four pregnant rats were exposed concurrently to 100-MHz CW RFR at 46 mW/sq cm at the Environmental Protection Agency, North Carolina (EPA) in a transverse electromagnetic (TEM) cell described in Smialowicz et al. (1981b). Mean SARs ranged from 2 W/kg for pregnant rats to 3 W/kg for neonates, with intermediate values for older pups. The exposures were performed during 0800-1400 daily on gestation day 6 through term, after which the offspring were exposed 4 hr daily for 97 days. Four other pregnant rats and their pups were similarly treated but sham-exposed. The treated offspring were delivered to GWU 14 months after cessation of exposure.

In all three experiments, the brain of each RFR-exposed animal was processed concurrently with its corresponding sham-exposed control. After anesthesia, each rat was fixed with formalin by cardiac perfusion and the cerebella were removed, divided in the midsagittal plane, embedded in paraffin, serially sectioned (10 micrometers thick), and stained. Six to nine parasagittal planes were selected and used as constant reference levels for matching cerebellar sections from RFR- and sham-exposed rats. Four to six serial sections were studied at each plane. Each section was projected, the boundaries of various regions were outlined, the geometric areas within the boundaries of interest were determined with a planimeter, and the Purkinje cells within the molecular-granular interface were counted. (Cells without visible nucleoli were not counted.) All cell counts were performed double blind by two individuals. Results were presented in terms of the mean density of Purkinje cells (number per sq mm) and standard error (SE) for the RFR- and sham-exposed groups.

In experiment 1, the pregnant rats delivered their pups within 12 hr after completion of their RFR- and sham-exposures and return to GWU. Three of the six pups exposed prenatally to RFR and three of the six pups sham-exposed prenatally were euthanized on postnatal day 1, and the remaining three pups of each group were euthanized 40 days after birth. The cerebella from the day-1 pups were not mature enough for clear discernment of the Purkinje cells, so no data were presented. The mean counts for the RFR-exposed 40-day pups were 35.87 +/- 1.71 cells/sq mm, as compared with 48.47 +/- 2.26 cells/sq mm for the sham-exposed 40-day pups. The 25.8% decrease for the RFR pups was significant (p<0.01).

In experiment 2, three pups exposed to RFR and three pups that were sham-exposed postpartum were euthanized immediately after return to GWU and the other three of each were euthanized 40 days after cessation of RFR- or sham-exposure. For the pups euthanized immediately, the mean counts were 87.46 +/- 4.62 cells/sq mm for the RFR-exposed pups and 115.68 +/- 5.30 cells/sq mm for the sham-exposed pups, a significant difference (p<0.01). For the 40-day pups, the mean counts were 31.61 +/- 1.22 and 33.98 +/- 1.05 cells/sq mm, respectively, a nonsignificant difference (0.05<p<0.1).

In experiment 3, the pups that were sham-exposed or exposed to 100-MHz RFR both in utero and postpartum were delivered to GWU 14 months after completion of the exposure regimen and were euthanized immediately. The mean counts were 24.74 +/- 0.71 and 28.33 +/- 1.21 for the RFR-exposed and sham-exposed pups, respectively, a significant difference (p<0.05).

The authors noted that the cerebellar rudiment and the Purkinje cells arise during the second half of gestation. They suggested that the significantly smaller Purkinje-cell counts obtained in experiments 1 and 3 (both of which involved exposure to RFR during that half of the prenatal period) could be due to reduction of the proliferative activity of the neuroepithelium by the RFR, or that the RFR could affect the migratory pattern of Purkinje cells and other microneurons in a manner that prevents the Purkinje cells from reaching the molecular-granular interface. The observation of smaller cell counts at both 40 days and 14 months after completion of RFR exposure was taken as indicating the permanence of this effect of prenatal exposure.

The results of experiment 2, in which significantly smaller Purkinje-cell counts were obtained for the pups exposed postnatally to RFR and euthanized immediately, and the nonsignificant difference in counts between the groups euthanized 40 days after RFR-and sham exposure, were taken as an indication of the reversibility of this effect. Presumably because brain cells are not regenerative, the authors suggested that postnatal exposure to RFR could temporarily affect the migration pattern of the Purkinje cells but not the proliferative activity of the external granular layer.

Although the positive results above appear to be statistically valid (from use of the t-test), they are questionable because the normalized SDs (ratio of SD to its mean value) for the RFR groups were virtually the same as for their respective sham-exposed groups. In view of the large variations in mean SAR at 2.45 GHz stated by the authors (0.8 to 6 W/kg), one would expect to observe some dose dependence of effect among the RFR-exposed animals that would be manifested as much larger values of normalized SDs than for the sham-exposed animals.

Albert et al. (1981b) conducted a similar study on squirrel monkeys that had been exposed perinatally elsewhere in an investigation of possible RFR-induced infant mortality described briefly in a note added in proof to a paper by Kaplan et al. (1982), discussed in Sections and In that infant-mortality study, Kaplan et al. (1982) exposed pregnant squirrel monkeys to 2.45-GHz amplitude-modulated RFR at a whole-body SAR of 3.4 W/kg (equivalent to plane-wave RFR at 10 mW/sq cm) 3 hr/day for 7 days/week, starting near the beginning of the first trimester of pregnancy. On parturition, each dam and its infant was exposed as a dyad for 6 months, and some offspring were exposed for 3 additional months without their dams. Albert et al. (1981b) obtained seven each of the RFR- and sham-exposed offspring used in that infant-mortality study.

These monkeys were weighed and anesthetized, and their brains were fixed with formalin by cardiac perfusion within 24 hr after receipt. The brains were removed and their weights and volumes were determined. The cerebella were dissected and the inferior vermis of each was separated and embedded in paraffin. The preparation was serially sectioned in the parasagittal plane at thicknesses of 10 micrometers and stained with hematoxylin-eosin. The uvula of the inferior vermis was selected for study because its mean cell density is representative of the mean cell density of the entire cerebellum.

Purkinje-cell density in each uvula was estimated by determining the area of the Purkinje cell layer and counting the number of cells therein by light microscopy, with the result expressed as the number of cells per sq mm. The linear density was also determined by counting the number of Purkinje cells per mm in the interface between the molecular and granular regions of each section and averaging the results over the sections. Only cells that displayed a visible nucleolus were counted. No significant differences between RFR- and sham-exposed groups were found in whole-body mass, brain mass, brain volume, or Purkinje-cell counts per sq mm or per mm.

These negative results are at variance with those of the previous study with rats (Albert et al., 1981a), in which the mean density of Purkinje cells in the cerebella of rats exposed prenatally to 2.45-GHz or 100-MHz RFR at whole-body SARs of about 2 W/kg was significantly lower than for sham-exposed rats. Albert et al. (1981b) suggested that differences in species and in exposure methods, geometrical configurations of the head, and exposure protocols might account for the differences in findings. It is possible that local SARs in the rat brain may be higher than in the squirrel-monkey brain at comparable whole-body SARs. On the other hand, organogenesis in the squirrel monkey is completed within the first trimester of pregnancy, during which the brain may be more sensitive to exogenous agents (such as RFR) than subsequently. However, perhaps the negative results of Albert et al. (1981b) render this point moot.

Merritt and Frazer (1975) conducted a study to determine whether RFR altered whole-brain levels of specific neurotransmitters in the mouse. They exposed male Swiss-Webster mice individually at 19 MHz (HF) for 10 min to either an electric (E) field or a magnetic (H) field in plastic ventilated cages placed within a near-field synthesizer (Greene, 1974), with the long body axes of the mice perpendicular to the E-field. The synthesizer was calibrated with near-field electric and magnetic probes (Greene, 1975). With 400 W input, the E-field was 6 kV/m and the H-field associated with this E-field was 6.4 A/m; the H-field was 41 A/m and the associated E-field was 2 kV/m. The authors noted that exposure at these levels did not increase rectal temperature, measured with a thermistor probe. Control mice were sham-exposed in the synthesizer.

Fifteen minutes after exposure, the mice were euthanized by brain-enzyme inactivation with a 2.45-GHz industrial microwave oven modified by the addition of a section of shorted waveguide; the head of each mouse was inserted through a hole in the waveguide section at the maximum E-field location. By test, brain temperature rose 40-50 deg C in 1 second. A group of sham-exposed mice was similarly euthanized (microwave controls) and another group of sham-exposed mice was euthanized conventionally by cervical dislocation (conventional controls). Following euthanization, the brains were quickly removed, immediately frozen in liquid nitrogen, and homogenized in cold acidified (10 mM HCl) butanol. Each brain was assayed by spectrofluorometry for serotonin (5HT) and its metabolite 5-hydroxyindole acetic acid (5HIAA), dopamine (DA) and its metabolite homovanillic acid (HVA), and norepinephrine (NE). The normalized whole-brain mean concentrations (in nanograms per gram) and SDs are shown in Table 32 (adapted from Table 1 of the paper), with the number of mice in each group in parentheses:


Treatment 5HT 5HIAAA DA HVA NE
E-field 745 780 1318 172 402
+/- SD 132 (5) 113 (5) 362 (5) 61 (5) 167 (5)
H-field 770 752 1487 182 533
+/- SD 79 (12) 62 (13) 374 (12) 44 (13) 123 (12)
Microwave Cont. 774 742 1355 184 468
+/- SD 50 (12) 121 (13) 287 (12) 62 (13) 88 (14)
Convent. Cont. 582 621 1025 184 367
+/- SD 86 (10) 75 (8) 108 (10) 62 (8) 38 (10)


By t-test, there were no statistically significant differences between the microwave controls and either E- or H-field-exposed animals in mean whole-brain concentrations for any of the neurotransmitters or their metabolites. However, the mean concentrations of all the neurochemicals except HVA were significantly higher (p<0.02) in the microwave controls than in the conventional controls. The authors suggested that these differences indicate that neurotransmitters were present having very rapid turnover and thus may be detectable after use of microwave- but not conventional brain-enzyme inactivation because of the shorter time required for the former.

The authors also noted that by frequency scaling, exposure of a mouse to 19-MHz RFR to obtain a specific whole-body SAR would require about 30-40 times higher incident power density than to obtain the same whole-body SAR in a human. On this basis, short-duration exposure of humans to 19-MHz E-fields up to about 150 V/m, or H-fields up to about 1 A/m may not have any detectable effects on the five neurochemicals studied.

Merritt et al. (1976) then sought to determine whether exposure to 1.6-GHz RFR would have any effect on the neurotransmitter concentrations in discrete regions of the rat brain. They inserted male Sprague-Dawley rats in individual ventilated cylindrical Plexiglas holders and placed each holder with its central axis upright and 43 cm along the axis of a horizontally radiating horn within an anechoic chamber. Exposures were for 10 min at 80 mW/sq cm. For hyperthermal controls, other rats were placed for 10 min in a chamber maintained at 78 deg C.

The mean rectal temperature rise in the rats exposed to the RFR was 4.09 +/- 0.55 (SE) deg C. The rise for the hyperthermal controls was 3.70 +/- 0.68 deg C. By scanning infrared thermography of a euthanized rat sectioned along a midsagittal plane and exposed at 200 mW/sq cm for 1 min, the temperature rise along the base of the brain was found to be comparable to the increase in core temperature in the intact rat.

The rats were euthanized by microwave brain-enzyme inactivation right after either treatment, the brains were quickly removed, and discrete regions were dissected out. Each brain region was then frozen in liquid nitrogen, weighed in the frozen state, homogenized in acidified butanol, and assayed for the following: norepinephrine, dopamine, serotonin, and 5-hydroxyindole acetic acid in the hypothalamus, corpus striatum, mid-brain, hippocampus, cerebellum, medulla, and cortex. Homovanillic acid, the metabolite of dopamine, was also assayed, but only in the corpus striatum because it cannot be detected in the other regions of the brain. The results were tabulated as mean concentrations (in ng/g) in each region, with SEs and numbers of determinations, and examined for statistical significance by t-test.

Regarding norepinephrine, the concentration in the hypothalamus of the RFR group was significantly lower (p<0.01) than in the control group, the only significant difference; the norepinephrine concentration in the hypothalamus of the hyperthermia group was 77% of that in the controls, but the difference was nonsignificant (p>0.05). Dopamine concentration in the corpus striatum of the RFR group was significantly lower (p<0.05) than in controls; it was also lower in the hypothalamus and labeled by the authors as significant (p<0.05), but a check of the p value showed it to be slightly larger than 0.05.

The concentrations of serotonin in the hippocampus of the RFR group and the cortex of the hyperthermia group were both significantly lower than in the controls (p<0.02 and p<0.05, respectively). All the differences in concentration of 5-hydroxyindole acetic acid in each brain region between RFR and control groups and between hyperthermia and control groups were nonsignificant. The mean homovanillic-acid concentrations in the corpus striata of the RFR- and hyperthermia groups were both significantly lower (p<0.05) than for the control group.

The authors concluded that: "the metabolism and/or transport of the putative neurotransmitters studied are altered by exposure of the rat to an environment that produces hyperthermia...The changes noted in the neurotransmitters in the brain areas were in the same direction for both the irradiated and hyperthermal animals. It then seems probable that these changes are a result of mechanisms involved in maintaining thermoregulation or a thermal effect on brain cells and not a direct effect of 1.6 GHz radiation on neural components."

Sanders et al. (1980) investigated the hypothesis whether exposure of brain tissue to RFR in vivo results in inhibition of the respiratory chain function followed by decreases in concentrations of adenosine triphosphate (ATP) and creatine phosphate (CP).

As indicated by the authors, the first entrance point to the electron transport chain, at which nicotinamide adenine dinucleotide (NAD) is reduced (by proton addition) to NADH, can be monitored continuously in situ by excitation at 366 nm and observation of NADH fluorescence at 460 nm with a time-sharing fluorimeter. Thus, if RFR imposes stresses on the cells that inhibit respiratory chain function or cell functions that utilize ATP and CP, the NADH level will increase (higher fluorescence); conversely, RFR-induced enhancement of the respiratory chain or increase of cell functions that utilize ATP and CP will reduce the NADH level.

Male Sprague-Dawley rats were anesthetized with sodium pentobarbital (40-45 mg/kg), the scalp and muscles at the side of the skull were removed, and a 4x8-mm aperture was made through the skull, leaving the dura intact. The head of the rat was rigidly held in place, the light from a 366-nm excitation source was focused to a spot (1-2 mm) on the cerebral cortex, and a fiber-optic bundle was directed toward the focal spot. The other end of the fiber-optic bundle was terminated at a wheel that housed a 460-nm filter for NADH-fluorescence measurements and a 366-nm filter for measuring reflectance from the surface.

Following preparation, the rat was exposed to 591-MHz CW RFR at 3.0 cm from a 3.0-cm-square dielectrically loaded horn (in the far field), with the electric vector parallel to the long axis of the rat. The radiation pattern of the horn was such that only the head of the rat was exposed. A grounded Faraday shield was used to avoid RFR interference with the fluorimeter electronics. Incident power densities at the exposure site were measured with a Raham radiation detection meter, and theoretical models were used to estimate SARs. The maximum normalized SARs at the surface of a 2.0-cm diameter sphere and a semi-infinite plane of brain tissue were given as 0.026 and 0.16 W/kg per mW/sq cm, respectively.

Baseline NADH fluorescence and reflectance levels were determined in one rat before exposure. When both baselines were steady for 5 min, RFR-exposure for 5 min at 13.8 mW/sq cm was begun. The 366-nm reflectance trace showed no significant deviation from baseline during exposure, but the NADH trace began to increase on initiation of the RFR, reached a maximum at 30 seconds, and showed a compensatory partial return toward baseline during the next 2.5 min, followed by a slow rise again during the remaining 2 min. The authors ascertained that the changes in NADH fluorescence observed in the baseline experiment were not artifactual by using exposures at 18 mW/sq cm and several layers of opaque cloth to block excitation light from reaching the brain.

Groups of rats given the same preexposure treatment as in the baseline experiment were sham-exposed or exposed at 13.8 mW/sq cm for 0.5, 1, 2, 3, or 5 min, during which NADH fluorescence was measured. Right after RFR-or sham-exposure, the head and neck were immersed in liquid nitrogen for at least 2 min, and the frozen head was removed and stored in liquid nitrogen. Later, the frozen cerebral hemisphere near the aperture in the skull was extracted and pulverized, and duplicate aliquots were assayed for ATP and CP. Groups of rats were sham-exposed or exposed at 5.0 mW/sq cm for 0.5 or 1 min and similarly processed.

As in the baseline experiment, NADH fluorescence increased at initiation of 5-min exposures at 13.8 mW/sq cm. The levels reached maxima ranging from 4% to 12.5% above baseline at 30 seconds, decreased slowly to minima of about 2% above baseline in the next 2.5 min, and then again increased slowly to 5% above baseline at 5 min. The concentrations of ATP and CP for each duration of sham-exposure and exposure to RFR at 13.8 mW/sq cm were graphed as percentages of baseline level. The sham-exposures did not yield any significant changes in either ATP or CP. For exposure to RFR, however, the ATP level dropped to a minimum of about 75% by 30 seconds and rose to levels that did not exceed about 90%; the CP level dropped to a minimum of about 60% by 30 seconds and rose to a maximum of about 85%. By Student's t-test, all differences in levels of each biochemical between RFR- and sham-exposure groups for corresponding durations were significant, most at the p<0.005 level. The authors also noted that the decreases in ATP and CP levels were correlated with the relative increases in NADH level.

The mean ATP and CP levels (in micromoles/g), SEs, and numbers of rats are presented in Tables 33 and 34 (adapted from Tables 2 and 3 of the paper) for exposure at 0 (sham), 5, and 13.8 mW/sq cm for 0.5 and 1 min:


Power Density 0.5-Min Exposure 1-Min Exposure
0 mW/sq cm 2.48 +/- 0.06 (11) 2.48 +/- 0.05 (11)
5 mW/sq cm 1.89 +/- 0.06 (6) 2.17 +/-0.07 (6)
13.8 mW/sq cm 1.93 +/- 0.17 (10) 2.19 +/- 0.09 (10)



Power Density 0.5-Min esposure 1-Min Exposure
0 mW/sq cm 3.32 +/- 0.14 (11) 3.27 +/- 0.15 (11)
5 mW/sq cm 1.99 +/- 0.16 (6) 2.17 +/- 0.31 (6)
13.8 mW/sq cm 2.17 +/- 0.30 (10) 2.11 +/- 0.31 (10)


Regarding these results, the authors stated: "Even the 5.0 mW/sq-cm were above threshold for decreasing brain ATP and CP levels. This does not imply that a threshold power level would not be found if the microwave exposure power level continued to be decreased." However, by t-test of the ATP results above, the decreases for exposure at 5 mW/sq cm for 0.5 and 1 min were highly significant (p<0.001 and p<0.01, respectively), but the differences between exposure at 5 and 13.8 mW/sq cm for either duration were nonsignificant (p>0.05). This was also true for the CP results. Both statistical findings appear to indicate the absence of dose-dependence, at least for exposure levels above the threshold postulated by the authors, or that a plateau had been reached.

Temperatures at 2-3 mm below the surface of the brain were also measured in this study. For this purpose, a thermistor was placed perpendicular to the electric vector immediately under the dura but on top of the brain, adjacent to the focal spot of the excitation light. Rats other than those used in the NADH, ATP, and CP assays were exposed at 0, 13.8, 18.0, 30.0, 40.0, and 47.0 mW/sq cm for 5 min. Rectal temperatures were also recorded. With sham-exposure, the heat sink due to the aperture in the skull was found to cause a decrease in brain temperature of 0.7 deg C (from 36.9 initially to 36.2 deg C at the end of the 5-min interval). For exposure at 13.8 mW/sq cm, the initial and final brain temperatures were 35.2 and 34.6 deg C, a decrease of 0.6 deg C. Decreases in brain temperature were also observed for exposure at 18.0 and 30.0 mW/sq cm, 0.5 and 0.1 deg C, respectively, but increases of 0.2 and 0.1 deg C were obtained at 40.0 and 47.0 mW/sq cm, respectively. Rectal temperatures remained constant at all exposure levels.

The authors concluded that because brain temperatures did not increase during RFR exposure at 5 or 13.8 mW/sq cm, the observed changes in NADH, ATP, and CP levels could not be ascribed to general tissue hyperthermia (but did not rule out local hyperthermia). Instead, the data support the hypothesis of RFR inhibition of electron transport chain function in brain mitochondria and results in decreased energy levels in the brain.

Subsequently, Sanders and Joines (1984) investigated the effects of hyperthermia alone and in conjunction with RFR. Male Sprague-Dawley rats were anesthetized with urethane (1250-1350 mg/kg ip) [instead of sodium pentobarbital] "to avoid barbiturate effects on brain energy metabolism." Rat preparation was similar to that in the previous study, except that the fiber-optic bundle was bifurcated near the measurement end; one arm served to convey 460-nm light from the brain to apparatus for measuring NADH fluorescence and the other arm conveyed 549-nm light to apparatus for measuring general brain fluorescence. Measurements of the latter were to permit correction of NADH fluorescence data for the effects of hemoglobin-content changes in the brain region studied, but no changes in 549-nm fluorescence were observed during and after RFR exposure, rendering such corrections unnecessary and showing that normal respiration rates were maintained in the rats during RFR exposure.

Groups of anesthetized rats were exposed to 591-MHz CW RFR for up to 5 min at 13.8 mW/sq cm. The upper and lower limits of spatial-average SAR in the brain were estimated in this paper to be 0.42 and 0.18 W/kg per mW/sq cm, based on prolate-spheroidal models having volumes of 110 and 25 cu cm in the E-polarization (Durney et al., 1978, pp. 94, 99), which yielded limits of 5.8 and 2.5 W/kg at 13.8 mW/sq cm. Also, temperatures were measured with a thermistor at a point about 2 mm below the surface of the brain during exposure at 60 and 100 mW/sq cm, and the rates of temperature increase were used (with 0.88 cal/g per deg C for brain-tissue specific heat) to estimate the local normalized SAR. The results for the two power densities were 0.613 and 0.626 W/kg per mW/sq cm, yielding a local SAR of about 8.5 W/kg for 13.8 mW/sq cm.

Other groups were subjected only to brain hyperthermia and still other groups to combined hyperthermia and RFR. In both sets of hyperthermia experiments, brain temperatures were held constant by use of heater pads and a temperature regulator. However, the method used to maintain brain temperatures constant was not described. At the end of the paper, the authors stated only that: "The temperature regulator was adjusted so that power would turn off when the selected temperature (35.6 or 37 or 39 deg C) was reached."

In an auxiliary experiment, ATP and CP assays in urethane-anesthetized RFR-exposed rats were compared with similar assays in pentobarbital-anesthetized RFR-exposed rats with the brains of both groups held at 35.6 deg C. The maximum deviation was 4.5%.

In the hyperthermia-only experiments, brain temperatures of the first 10 urethane-anesthetized, sham-exposed rats were maintained at 35.6 deg C. The assays of ATP and CP yielded mean concentrations of 2.47 +/- 0.03 (SE) and 3.44 +/- 0.11 micromoles/g, respectively. As indicated by the authors, these values did not differ significantly from those reported by Sanders et al. (1980) (for 0 mW/sq cm in Tables 33 and 34). The authors then sham-exposed groups with brain temperatures maintained at 37.0, 39.0, or 41.0 deg C and plotted the mean ATP and CP concentrations vs brain temperature as percentages of the concentrations at 35.6 deg C. The percentages of both ATP and CP declined monotonically with increases in brain temperature, with the rate of decline for CP higher. At 39 deg C, ATP and CP decreased to about 90% and 70%, respectively; at 41 deg C, the decreases were to about 70% and 45%, respectively.

Rats with brain temperatures held at 35.6 deg C were exposed for 0.5, 1, 2, 3, or 5 min at 13.8 mW/sq cm, and the percentage increase in NADH and mean percentages of ATP and CP (relative to the concentrations at 35.6 deg C for sham-exposed rats) were plotted vs exposure duration. These results, displayed in Fig. 3 of the paper, were evidently the same as those shown in Fig. 1 of Sanders et al. (1980), i.e., ATP decreased to about 75% and CP to about 60% in the first half-min. A similar set of experiments was performed with brain temperatures held at 39.0 deg C (RFR plus hyperthermia). At 0 min, the ATP and CP levels respectively were about 90% and 70%, decreases ascribed to the hyperthermia, and the levels declined further to minima of about 60% and 45% by 1 min of exposure. By 5 min, the ATP level rose linearly to about 80% and the CP level rose nonmonotonically to about 50%.

Agroup of sham-exposed rats with brain temperatures at 35.6 deg C was rendered hypoxic by giving them air consisting of 2% oxygen and 98% nitrogen for 0.5, 1, or 2 min. At 0.5 and 1 min, the mean ATP level of these rats remained 100% but dropped to about 85% at 2 min; by contrast, the mean CP level dropped to 80% and 60% at 0.5 and 1 min, and further to about 25% at 2 min. The authors stated: "Note that in the hypoxic animals brain ATP did not decrease until CP decreased below 59% of control levels. This illustrated the functional role of CP as a support system for maintaining ATP within narrow limits. Brain CP will decrease to levels of approximately 55-60% of control levels at normothermic conditions before any significant decrease in ATP can be observed."

The general conclusion was: "The decreases in ATP and CP in the 39 deg C brain during microwave exposure were significant and resulted in ATP and CP being much lower than observed at 35.6 deg C. Thus, at 39 deg C when the brain metabolic rate was increased, subsequent microwave exposure rapidly induced further decreases in ATP and CP, similar to the 35.6 deg C microwave exposure data, ie, without a further increase in brain temperature; [these results] are consistent with the concept of direct microwave inhibition of energy metabolism." The latter statement seems open to question, however, because with an estimated local SAR of 8.5 W/kg at 13.8 mW/sq cm, considerable heating must have occurred.

In another paper, Sanders et al. (1984) described the results of similar experiments, but with frequencies of 200 MHz and 2.45 GHz in addition to 591 MHz. In this study, rats (surgically prepared as in the previous studies) were exposed to RFR in a horizontal microstrip transmission line system having a 6-cm plate separation, with the anesthetized rat placed on the ground plane. Precautions were taken to ensure that RFR-induced artifacts in NADH fluorescence were negligible. Incident power densities at the position of the rat's head (in the absence of the rat) were measured with a Raham radiation hazard meter.

Local SARs at a depth of 2-3 mm below the top surface of the brain of a dead rat were determined from measurements of temperature rise vs time with a Vitek isotropic probe during exposure of the carcass to each frequency at 60 and 100 mW/sq cm. The normalized SARs at 200, 591, and 2450 MHz were 0.046, 0.185, and 0.368 W/kg per mW/sq cm.

Brain temperature at the same depth was measured with the Vitek probe in urethane-anesthetized rats. The authors noted that the concentrations of ATP and CP did not significantly change when brain temperature varied by 0.4 deg C (Sanders and Joines, 1984) and that urethane anesthetic lowers the rat's body temperature to about 35.5 deg C. They stated: "Consequently, the brain temperature in these experiments was maintained at 35.6 +/- 0.3 deg C using a hot air blower which was directed into the microstrip toward the animal. The blower cycled on and off when the brain temperature reached 35.3 and 35.8 deg C, respectively. Occasionally, the brain temperature would increase to 35.9 deg C. Without the blower, we found that the brain temperature in sham-exposed animals would decrease from 35.6 deg C to 35.1 deg C in 5 min. In RF-exposed animals the brain temperature did not decrease as rapidly and the blower ran less often. On average, the temperature of the brain in RF-exposed animals was somewhat lower than in sham-exposed animals, ie, in the RF-exposed animals the brain temperature was between 35.4 and 35.6 deg C much of the time."

The authors indicated that brain fluorescence varied widely from rat to rat because of differences in brain vasculature and noted: "Therefore, the fluorescence data for different animals should not be averaged; each animal must be used as its own control." For this reason, the percent changes in NADH level for six individual rats exposed to 200-MHz RFR at power densities in the range 0.5-40.0 mW/sq cm (0.02-1.84 W/kg) and for six other rats exposed to 591-MHz RFR in the same power-density range (0.09-7.40 W/kg) were tabulated in the paper. At 200 and 591 MHz, there was a dose-response relationship for each rat in the percentage change in NADH for levels up to 10 mW/sq cm (0.46 and 1.85 W/kg, respectively). Above this power density, the NADH responses showed plateaus. At 2.45 GHz, significant NADH changes were not observed, leading the authors to suggest that the effect was frequency-dependent.

The authors also suggested that the important exposure parameter may be the electric field intensity in brain tissue, and using the dielectric constant and conductivity for each frequency, they converted incident power densities to tissue field strengths, obtaining 12.5, 22.5, and 20.7 V/m per mW/sq cm for 200, 591, and 2450 MHz, respectively. In addition, by extrapolating the 200- and 591-MHz curves for maximum and minimum NADH change to zero (the abscissa where the two curves met), they estimated that the threshold field for effect was 3-5 V/m.

The curves for percent NADH increase and percentage changes in ATP and CP vs duration for exposure to 591 MHz at 13.8 mW/sq cm (2.6 W/kg, 311 V/m) were stated to be the same as those given in Fig. 1 of Sanders et al. (1980). For 200 MHz (0.63 W/kg, 173 V/m), the NADH fluorescence curve and percentage concentrations of ATP were qualitatively similar to those for 591 MHz, but the changes were smaller: at 0.5 min, the NADH fluorescence increased to a maximum only about 5% above controls and ATP concentration dropped to about 80% of controls; however, the changes in CP concentration were nonsignificant for all durations (up to 5 min). For 2.45 GHz (5.1 W/kg, 286 V/m), no changes were seen in NADH level (as noted above) and the percentage changes in ATP and CP concentrations were statistically nonsignificant.

Last, rats were exposed to 591-MHz RFR at 13.8 mW/sq cm for durations up to 20 min. The data for durations up to 5 min were the same as those in Fig. 1 of Sanders et al. (1980). Beyond 5 min, the NADH level continued to slowly rise linearly, to a level 10% above baseline at 20 min (about equal to the maximum level at 0.5 min). The ATP concentration continued to decrease linearly, from 88% of controls at 5 min to 65% at 20 min. The CP concentration, 65% of controls at 5 min, dropped to a minimum of 45% at 10 min and rose to 60% at 20 min.

In their discussion the authors noted: "Adenosine triphosphate [ATP] is a key compound in energy metabolism because it is the carrier of energy to the processes in living cells. The relationship of ATP to CP and NADH in the metabolic pathway of energy production has been described in detail elsewhere (Siesjo, 1978). Briefly, NADH is oxidized to produce ATP in the mitochondria. In addition, brain ATP concentration is maintained at the expense of CP. Siesjo (1978) has shown in the hypoxic rat that brain ATP was maintained at control levels when the CP concentration had decreased to 62.4% of control CP levels. Also, when the CP concentration had decreased to 35.3% of control levels, brain ATP had decreased to only 77.8% of control levels. Therefore, when demand for ATP is higher than the mitochondrial production capacity, CP is rapidly converted to ATP by CP-kinase to sustain ATP levels, and significant decreases in CP levels are observed prior to any decrease in ATP." Thus, in interpreting their results, the authors stated:

"In the experiments presented here, increased NADH levels suggest an inhibition of mitochondrial oxidative phosphorylation. This reversible effect occurs during irradiation at 200 MHz and 591 MHz but not at 2,450 MHz...at 591 MHz the CP concentration drops more rapidly than the ATP concentration, which is expected. However, the ATP concentration should be maintained until the CP concentration drops below 60% of normal levels. This is clearly not the case here, where the microwave exposures resulted in a decrease in ATP levels to 75% of controls when CP levels are no lower than 60% of controls. During the 200-MHz exposures, brain ATP levels decreased to 80-90% of controls even though CP levels were not significantly decreased. These results are not consistent with normal energy metabolism where ATP levels are maintained at the expense of CP, and suggest that at 200 MHz there is RF inhibition of the CP-kinase reaction converting CP to ATP."

In still another study, Sanders et al. (1985) compared the effects of CW, sinusoidal-amplitude-modulated, and pulsed 591-MHz RFR on NADH fluorescence and concentrations of ATP and CP in the brains of rats prepared as in the previous studies and exposed in the microstrip system. The initial temperature in the brain of the anesthetized rat was 35.6 deg C and decreased by 0.27 deg C during 5-min exposure at 13.8 mW/sq cm. No increase in brain temperature was seen during any of the RFR exposures. As in Sanders et al. (1984), because of the brain vascularity differences among rats, each rat served as its own control for NADH fluorescence, i.e., changes were expressed as percentages of its preexposure (baseline) NADH level.

In the sinusoidal-modulation experiments, the modulation was essentially 100% at frequencies ranging from 4 to 32 Hz in 4-Hz increments, and the exposures were for 5 min at average power densities of 10 and 20 mW/sq cm (1.8 and 3.6 W/kg). Mean percentage increases in NADH fluorescence vs modulation frequency were plotted (with SE bars) separately for the two RFR levels. At 20 mW/sq cm, NADH fluorescence was about 9% higher than baseline for modulations at 4 and 8 Hz, it increased to a maximum of 11% at 24 Hz, and decreased to slightly less than 9% at 32 Hz. The changes at 10 mW/sq cm were qualitatively similar but smaller, ranging from about 4.5% above baseline at 4 Hz to a relatively flat maximum of about 5.5% at 16 Hz. The SEs at both RFR levels were about 10% of the means. By analysis of variance (ANOVA) and comparison of treatment to control, the effect of exposure and difference in exposure levels were both statistically significant (p<0.001). However, although the data on modulation frequencies appeared to indicate a trend, the difference across modulation frequency was nonsignificant (p>0.1).

In the pulsed-RFR experiments, the pulses used were 5 microseconds in duration at 250 or 500 pps and the average power densities ranged from 0.5 to 13.8 mW/sq cm (0.09 to 2.5 W/kg). The percentage increases in NADH fluorescence were plotted vs average power density for each of four rats given 500 pps. At 13.8 mW/sq cm (2.5 W/kg), the largest and smallest increases in NADH fluorescence were about 8% and 4%, but the thresholds for all four rats were within the range 0.4-0.5 mW/sq cm (0.07-0.09 W/kg). Also shown was the curve for one of two rats given 250 pps. At 13.8 mW/sq cm, its increase in NADH fluorescence was about 5%, but its threshold was within the range 1.8-1.9 mW/sq cm (0.33-0.34 W/kg), about fourfold higher than for the rats given 500 pps. Analyses of variance of the NADH data for rats given 250 and 500 pps (six each) indicated a significant effect of exposure for each type of modulation.

The effects of CW, 16-Hz amplitude-modulated, and pulsed (500 pps) 591-GHz RFR, all at 13.8 mW/sq cm (2.5 W/kg) for durations of 0 (sham), 0.5, 1, and 5 min on the mean ATP and CP concentrations were tabulated for comparison. Multivariate ANOVA indicated a highly significant effect of exposure (p<0.001, Hotelling-Lawley) on the joint ATP-CP response and that the partial correlation between ATP and CP was high (p<0.001). The authors noted that this correlation was expected because the function of the CP-creatine phosphokinase system is to maintain ATP levels. The ATP and CP levels of all treated rats except one differed significantly from the levels for sham-exposed rats (p<0.05, Dunnett's test). The ATP level for one rat exposed to the pulsed RFR for 5 min was the exception. The authors stated: "The significance levels associated with the CP data should be viewed with some caution, however, since Bartlett's test showed unusual variability in the standard deviations."

From the tabulated data, the mean ATP or CP concentrations for exposure to CW, amplitude-modulated, and pulsed RFR did not differ significantly for corresponding durations.

The authors stated in their discussion: "The change in brain temperature was found to be -0.1 to -.04 deg C compared to preexposure temperatures during all exposures; ie, from the initial temperature of 35.6 deg C the final brain temperature was in the range of 35.2 to 35.5 deg C after a 5-min exposure. Thus the observed changes in brain NADH fluorescence and ATP and CP concentrations cannot be attributed to microwave-induced brain hyperthermia. The changes are consistent with the hypothesis of direct microwave inhibition of ATP production in the mitochondria, possibly by microwave-induced molecular dipole oscillations in the divalent metal-containing enzymes and/or electron transfer sites; such action could compromise the conformation and/or stereospecificity requirements for normal function."

In overall conclusion, RFR can cause observable histopathological and histochemical changes in the central nervous system of animals, but most of the positive findings were evidently thermally induced. The studies by Sanders and coworkers, however, are of considerable interest because their positive results were obtained in the absence of measurable tissue hyperthermia. The significance of their findings (direct-RFR-inhibition of respiratory chain function) with regard to possible hazards to human health is not clear at present.


Albert, E.N. and M. DeSantis
Ann. N.Y. Acad. Sci., Vol. 247, pp. 87-108 (1975)

Albert, E.N. and M. DeSantis
In C.C. Johnson and M.L. Shore (eds.), BIOLOGICAL EFFECTS OF ELECTROMAGNETIC WAVES, U.S. Department of Health, Education, and Welfare, HEW Publication (FDA) 77-8010, pp. 299-310 (1976)

Albert, E.N., M.F. Sherif, N.J. Papadopoulos, F.J. Slaby, and J. Monahan
Bioelectromagnetics, Vol. 2, No. 3, pp. 247-257 (1981a)

Albert, E.N., M.F. Sherif, and N.J. Papadopoulos
Bioelectromagnetics, Vol. 2, No. 3, pp. 241-246 (1981b)

Chou, C.-K. and A.W. Guy
IEEE Trans. Microwave Theory Tech., Vol. 26, No. 3, pp. 141-147 (1978)

Courtney, K.R., J.C. Lin, A.W. Guy, and C.-K. Chou
IEEE Trans. Microwave Theory Tech., Vol. 23, No. 10, pp. 809-813 (1975)

Durney, C.H., C.C. Johnson, P.W. Barber, H.W. Massoudi, M.F. Iskander, J.L. Lords, D.K. Ryser, S.J. Allen, and J.C. Mitchell
USAF School of Aerospace Medicine, Brooks AFB, TX, Report SAM-TR-78-22 (1978)

Frey, A.H. and E. Seifert
Life Sci., Vol. 7, No. 10, Part II, pp. 505-512 (1968)

Frey, A.H.
IEEE Trans. Microwave Theory Tech., Vol. 19, No. 2, pp. 153-163 (1971)

Galvin, M.J., D.L. Parks, and D.I. McRee
Radiat. Environ. Biophys., Vol 19, pp. 149-156 (1981c)

Greene, F.M.
U.S. Department of Commerce, National Bureau of Standards, NBS Technical Note 652 (1974)

Greene, F.M.
U.S. Department of Commerce, National Bureau of Standards, NBS Technical Note 658 (1975)

Kamenskii, Yu.I.
Biophys., Vol. 9, No. 6, pp. 758-764 (1964)

Kamenskii, Yu.I.
In SOCIETY OF NATURALISTS, Moscow, Vol. 28, pp. 164-172 (Engl. Trans., 1968)

Kaplan, J., P. Polson, C. Rebert, K. Lunan, and M. Gage
Radio Sci., Vol. 17, No. 5S, pp. 135-144 (1982)

McRee, D.I. and H. Wachtel
Radiat. Res., Vol. 82, pp. 536-546 (1980)

McRee, D.I. and H. Wachtel
Radiat. Res., Vol. 91, pp. 212-218 (1982)

Merritt, J.H. and J.W. Frazer
USAF School of Aerospace Medicine, Brooks AFB, Texas, Report SAM-TR-75-28 (August 1975)

Merritt, J.H., R.H. Hartzell, and J.W. Frazer
USAF School of Aerospace Medicine, Brooks AFB, Texas, Report SAM-TR-76-3 (February 1976)

Millar, D.B., J.P. Christopher, J. Hunter, and S.S. Yeandle
Bioelectromagnetics, Vol. 5, No. 2, pp. 165-172 (1984)

Olcerst, R.B. and J.R. Rabinowitz
Radiat. Environ. Biophys., Vol 15, pp. 289-295 (1978)

Portela, A., et al.

Rothmeier, J.

In THE NERVOUS SYSTEM AND ELECTRIC CURRENTS, Plenum Press, N.Y., Vol. 1, pp. 57-69 (1970)
Sanders, A.P., D.J. Schaefer, and W.T. Joines

Bioelectromagnetics, Vol. 1, No. 2, pp. 171-181 (1980)

Sanders, A.P. and W.T. Joines
Bioelectromagnetics, Vol. 5, No. 1, pp. 63-70 (1984)

Sanders, A.P., W.T. Joines, and J.W. Allis
Bioelectromagnetics, Vol. 5, No. 4, pp. 419-433 (1984)

Sanders, A.P., W.T. Joines, and J.W. Allis
Bioelectromagnetics, Vol. 6, No. 1, pp. 89-97 (1985)

Shtverak, I., K. Marha, and G. Pafkova
In P. Czerski et al. (eds.), BIOLOGIC EFFECTS AND HEALTH HAZARDS OF MICROWAVE RADIATION, Polish Medical Publishers, Warsaw, pp. 141-144 (1974)

Siesjo, B.K.
John Wiley and Sons, New York (1978)

Smialowicz, R.J., J.S. Ali, E. Berman, S.J. Bursian, J.B. Kinn, C.G. Liddle, L.W. Reiter, and C.M. Weil
Radiat. Res., Vol. 86, pp. 488-505 (1981b)

Tinney, C.E., J.L. Lords, and C.H. Durney
IEEE Trans. Microwave Theory Tech., Vol. 24, No. 1, pp. 18-24 (1976)

Tolgskaya, M.S. and Z.V. Gordon
(Translated from the original Russian text published by Meditsina Press, Moscow, 1971), Consultants Bureau, New York-London (1973)

Wachtel, H., R. Seaman, and W. Joines
Ann. N.Y. Acad. Sci., Vol. 247, pp. 46-62 (1975)