Figure ?. 2. Cosmogenic neutron spectrum measured in a shipping container in a building in Livermore, CA, (green curve) and the spectrum calculated by LANL using MNP6 and a detailed model of the building and container (black curve). The three broad peaks and the slowing-down plateau region typical of terrestrial cosmic-ray neutron energy spectra are labeled.
Figure ?.3 shows cosmogenic neutron spectra measured in a shipping container placed in different locations on container ships and a calculated neutron spectrum from a hypothetical model of shielded weapons-grade plutonium (dashed curve). The red curve was measured with the container housing the neutron spectrometer located on the steel deck of the ship with empty steel containers on top of it. The blue curve was measured with the spectrometer container perched on top of the stack of full containers, shielded from the steel of the ship. (These spectra are the same as the red and blue spectra shown in Figure ?.1.) The neutron flux of the red spectrum from the evaporation peak down to the thermal region is about a factor of two higher than for the blue spectrum. This difference is caused by the high-energy cosmogenic neutrons striking the relatively large iron nuclei in the steel of the ship and containers, generating more evaporation neutrons than they do in the smaller nuclei of lower atomic weight materials. The excess of neutron flux in the energy region where search detectors are sensitive is known as the neutron “ship effect.” Most of the enhanced-background region of the background spectrum overlaps the spectrum from shielded plutonium fission.
Figure ?. 3. Cosmogenic neutron spectra measured in a shipping container placed in different locations on container ships (red and blue curves) and a calculated neutron spectrum from a hypothetical model of shielded weapons-grade plutonium (dashed curve).
In the currently released version, MCNP6.1.1, solar modulation was done using year-averaged solar modulation potentials between 1960 and 2005 determined from neutron monitor data using the method of McKinney et al. (2006). Within that date range, the solar modulation potential was interpolated, and beyond 2005 it was predicted using a sinusoidal fit over several solar cycles. Future versions of MCNP6 will use solar modulation potentials from Usoskin et al. (2011) determined from neutron monitor and ionization chamber data between 1936 and the most recent past year. Beyond the date of the latest included data, the modulation potential is determined by an optimized sinusoidal fit over the last 8 to 14 years that include the two most recent extrema (Liegey et al. 2016). In principle, current daily or even hourly values of the solar modulation parameter could be used, but that can only be done if there is an operating network of neutron monitors with readily accessible real-time online data.
The possible improvement in accuracy that could be gained by using short-term real-time neutron monitor data to determine solar modulation for cosmogenic neutron flux calculations is shown in Figure ?.4 The figure shows graphs of monthly and daily averages of the count rate in the neutron monitor in Newark, Delaware relative to the long term average from 1967 to 2005. Even for this sea-level neutron monitor on the mid-Atlantic coast, the daily average count rate is often several percent different from the monthly average and it is not rare for the daily average neutron flux to be as much as 10% different from the monthly average.
Figure ?. 4. Relative count rate of the neutron monitor in Newark, Delaware. The green curve shows daily average count rates; the blue curve shows monthly averages.
Neutron detection and cosmogenic background neutrons are important not only for detecting hidden nuclear threats within and at the borders of the U.S. homeland, but also for nuclear treaty verification. Accurate knowledge of the cosmogenic neutron flux at a given place and time could improve treaty verification in several ways, all of which would depend on having reliable real-time or recent neutron monitor data.
A neutron detector called simply “radiation detection equipment” (RDE) was introduced under the Intermediate-Range Nuclear Forces (INF) Treaty between the U.S. and the Soviet Union and later used under the Strategic Arms Reduction Treaty (START) between the U.S. and Russia for on-site inspections to determine the absence or presence and number of nuclear warheads. The RDE is an array of helium-3 proportional counters surrounded by polyethylene, and it was designed to efficiently detect neutrons from fission (McNeilly and Rothstein 1994). At inspection sites, the proper operation and calibration of the RDE was determined using an americium-lithium (Am-Li) neutron source brought with the inspection teams, and the background count rate was separately measured. Transport of the neutron source in its shielding was awkward, and is more so now that radiation detection is employed at the borders that inspection teams must cross. For possible use in future treaties, it has been proposed that proper operation and calibration could be determined using the background count rate instead of the Am-Li neutron source. This can only be done if the cosmogenic neutron flux and energy spectrum is accurately known at the location of the inspection site at the time of the inspection. This could be achieved by experimentally verified calculations like the LANL calculations described above, but it would require real-time neutron monitor data. To be acceptable to each treaty participant, each country would want data from neutron monitors it can confidently rely upon.
The U.S. is not a signatory of the Comprehensive Test Ban Treaty (CTBT), but our country has an interest in ensuring its provisions can be verified. Protocol Part II of the CTBT (Comprehensive Test Ban Treaty Organization 2016) allows for on-site inspections to detect if an underground nuclear test has occurred. Part of such an inspection would include detecting radioactive gasses produced in the ground by fission neutrons. Along with radioactive xenon, radioargon isotopes, particularly 37Ar, are being considered. To understand soil air measurements taken during an on-site inspection, the radioargon background due to cosmic-ray-induced activation must be understood. Johnson et al. (2015) have used the cosmic-ray source feature of MCNP6 to calculate the cosmogenic neutron flux at ground level as a function of date during the solar magnetic activity cycle, latitude of sampling location, geology of the sampling location, and sampling depth. After the cosmic neutron flux was obtained, the rate of radioargon production was calculated. Radioargon production was shown to be highly dependent on the soil composition, particularly calcium content, as well as on latitude and solar magnetic activity. The half-life of 37Ar is 35.04 days. In a real on-site inspection, the background 37Ar level would depend on the cosmogenic neutron fluence over the past few months, so determining it accurately would require neutron monitor data for that period.
Some of the uses of cosmogenic neutron flux determinations using neutron monitor data for homeland and national security applications have been described above. Without ongoing reliable neutron monitor operation, efforts to improve detection of terrorist nuclear threats and nuclear treaty verification will be significantly impaired.
Comprehensive Test Ban Treaty Organization (2016) “Treaty text.” https://www.ctbto.org/the-treaty/treaty-text/.
Goldhagen, P. (2000), “Overview of aircraft radiation exposure and recent ER-2 measurements.” Health Phys.79: 526-544.
Goldhagen, P., J. M. Clem, and J. W. Wilson (2004), “The energy spectrum of cosmic-ray induced neutrons measured on an airplane over a wide range of altitude and latitude.” Radiat. Prot. Dosim.110: 387-392.
Goldhagen, P. (2011), “An extended-range multisphere neutron spectrometer with high sensitivity and improved resolution.” Nucl. Technol.175: 81-88.
Gordon, M. S., P. Goldhagen, K. P. Rodbell, T. H. Zabel, H. H. K. Tang, J. M. Clem, and P. Bailey (2004), "Measurement of the flux and energy spectrum of cosmic-ray induced neutrons on the ground," IEEE Trans. Nucl. Sci.51: 3427-3434.
Johnson, C., H. Armstrong, W. H. Wilson, and S. R. Biegalski (2015), “Examination of radioargon production by cosmic neutron interactions.” J. Env. Radioactivity 140: 123-129.
Liegey, L. R., J. R. Tutt, T. A. Wilcox, and G. W. McKinney (2016), “Predicting future solar modulation and implementation in MCNP6.” LANL report LA-UR-16-20290.
McKinney, G. W., D. J. Lawrence, T. H. Prettyman, R. C. Elphic, W. C. Feldman, and J. J. Hagerty (2006), MCNPX benchmark for cosmic ray interactions with the Moon, J. Geophys. Res., 111: E06004, doi:10.1029/2005JE002551.
McKinney, G. W., H. J. Armstrong, M. R. James, J. M. Clem, and P. Goldhagen (2012), “MCNP6 Cosmic-Source Option,” Proceedings of ANS Annual Meeting, Chicago, IL, June 24-28, 2012, LANL report LA-UR-12-00196.
McMath, G. E., G. W. McKinney, and T. Wilcox (2014), “MCNP6 Cosmic & Terrestrial Background Particle Fluxes – Release 4.” Proceedings of ANS Annual Meeting, Reno, NV, June 15-19, 2014, LANL report LA-UR-14-20090.
McMath, G. E. and G. W. McKinney (2014), “MCNP6 Elevation Scaling of Cosmic Ray Backgrounds.” ANS RPSD 2014 - 18th Topical Meeting of the Radiation Protection & Shielding Division of ANS Knoxville, TN, September 14 – 18, 2014, on CD-ROM, American Nuclear Society, LaGrange Park, IL.
McNeilly, J. H. and B. D. Rothstein (1994), “Radiation detection equipment (RDE) comparative evaluation test program volume 1—point source measurements.” Defense Nuclear Agency Technical Report DNA-TR-93-160-V1, downloaded from http://www.dtic.mil/dtic/tr/fulltext/u2/a283003.pdf.
Pelowitz, D. B., A. J. Fallgren, and G. E. McMath, editors (2014), “MCNP6 User’s Manual, Code Version 6.1.1 beta, June 2014, Manual Rev. 0” LANL report LA-CP-14-00745, Rev. 0. Export controlled; distribution authorized to U.S. Government agencies and their contractors.
Usoskin, I. G., G. A. Bazilevskaya, and G. A. Kovaltsov (2011), “Solar modulation parameter for cosmic rays since 1936 reconstructed from ground-based neutron monitors and ionization chambers, J. Geophys. Res., 116: A02104. [Data after year 2009 available at http://cosmicrays.oulu.fi/phi/Phi_mon.txt .]
Neutron Monitor Uses in Aviation Radiation Safety
Contributed by: Kyle Copeland, Ph.D.*
*U.S. Federal Aviation Administration, Civil Aerospace Medical Institute, Aerospace Medical Research Division, Protection and Survival Research Laboratory, Numerical Sciences Research Team. Mail Route AAM-630, 6500 S. MacArthur Blvd. Oklahoma City, OK 73169, USA.
Ionizing radiation exposure is an unavoidable part of daily life. Even our bodies are radioactive, due to the radioactive chemicals from which we are made. Ionizing radiation dose in humans is usually reported in gray (Gy) or sievert (Sv). The gray is the unit of absorbed dose, the average energy per unit mass deposited by the radiation. Deterministic effects of radiation exposure, i.e., those effects which are caused by cell killing of large numbers of cells, are traditionally considered in terms of gray. The sievert is the name used when the absorbed dose is weighted for health effects which are probabilistic in nature, i.e., stochastic effects. This weighting can be based on type of radiation, the tissue being irradiated, or both. The most common use of sievert is the quantity effective dose, which attempts to weight all radiation exposures with respect to cancer inductions, genetic effects, and length of life lost, and quality of remaining life (ICRP, 2007).
The global average dose per person from natural sources is about 2.4 mSv (or 2400 Sv: 1000 Sv = 1 mSv) each year, just from living on Earth, with most of the dose coming from radon exposure; for the U.S., the estimate is 3.1 mSv (United Nations Scientific Committee on the Effects of Atomic Radiation [UNSCEAR], 2000; National Council on Radiation Protection and Measurements [NCRP], 2009). Those living at unusually high altitudes or other places with unique living conditions (e.g., in the mountains, a coral atoll, etc.) can get a few times more or less than the average dose. Most of the natural variation is the result of shielding afforded by the Earth's atmosphere to galactic cosmic radiation (GCR) and variation in local radon levels, but radioactivity in the soil can also be very important.
While their doses from manmade sources are low, aircrews are among the highest occupationally exposed populations. Both the NCRP and UNSCEAR rank aircrew as among the most exposed occupations, at 3 mSv per year, almost all from GCR, roughly equaling their off-duty natural background exposure (UNSCEAR, 2000; NCRP, 2009). As aircraft cruise altitudes continue to increase, this dose will rise. According to the U.S. Bureau of Labor Statistics (BLS), there were about 210,000 crewmembers (pilots and flight attendants) in the U.S. labor force in 2014 (BLS, 2016).
There are several potential sources of radiation in aviation, but most of it usually comes from GCR. A decent approximation for USA based commercial flights within the Northern hemisphere of a few hours or more is about 5 Sv per hour of flight time. Doses per hour on shorter flights and on flights away from the poles, such as trans-equatorial flights, are lower. Equivalently scaled doses from other sources are extremely rare:
While a nuclear accident or spillage of radioactive cargo could cause a large exposure to aircrew or passengers. Such nuclear accidents are very rare and few flights carry radioactive cargo. Radioactive cargo is transported in well-shielded packages spaced far apart from each other to minimize exposure to aircraft occupants. Estimated annual dose: < 0.13 mSv per year (Warner-Jones et al., 2003).
Solar ionizing radiation is almost always orders of magnitude weaker than galactic cosmic radiation, the exception being during rare, intense solar particle events (SPEs). Because the even the most intense solar particle spectrum is much softer than the GCR spectrum, they typically only significantly increase doses on high-latitude flights where the ions with rigidities of a few GV or less are permitted access to the atmosphere. A study of the SPEs that occurred from 1986-2007 by Copeland et al. (2008) found that even at 18 km altitude, solar proton events with dose rates at high latitudes that exceed 20 Sv/h have occurred less than once per year, on average. The events roughly followed solar activity in frequency. Events resulting in possible in-flight exposures exceeding the 1 mSv recommended limit for pregnant crewmembers at 12 km (40,000 ft.) and above are extremely rare, occurring about once a decade.
Lightning and terrestrial gamma-ray flashes (TGFs) are associated with large thunderstorms, which aircraft avoid if it is possible. About 1 in 1000 flights is struck by lightning, mostly at low altitudes, such as during approach maneuvers. While a dose of up to 30 mSv has been postulated (Dwyer et al., 2010) for a worst case exposure from a TGF, they are associated with only a small fraction of lightning strokes (best evidence is for about 1 in 1000 or less) and are thought to occur only above about 10-15 km. Most lightning only produces insignificant amounts of soft X-rays.
GCR is relatively constant. It cannot be practically shielded against and effective dose increases as altitude increases. At the levels of exposure found in commercial aviation today, a few millisieverts of occupational exposure per year is a reasonable estimate of occupational exposure for crewmembers.
As indicated by the threshold doses in Table 1, the potential exposures from all the exposures listed above are well below the dose levels that can cause deterministic effects. It is the stochastic effects (Table 2) that are of concern. Though the odds are low, cancer induction is the most likely effect. It should also be noted that these risk estimates are for large populations, not individuals. There is too much variation in sensitivity and too little knowledge of cancer induction pathways to estimate risk for any particular individual.
Table 1. Low Dose Deterministic Effects
Risks to conceptus (mental retardation, malformation, etc.)
Transient mild nausea and headache in adults
* Assumes acute exposures, 1 Gy here is equivalent to 1 Sv in Table 2.
Genetic defect in first or second generation (child or grandchild) following irradiation before conception
0.4 in 100,000 per mSv
2.4 in 1,000,000 per mSv
Cancer (non-fatal or fatal)
34 in 100,000 per mSv
23 in 100,000 per mSv
Cancer (fatal only)
8.0 in 100,000 per mSv
6.3 in 100,000 per mSv
*Risks assume exposure to high-LET radiation (i.e., no DDREF) (ICRP, 2008).
Table 3. Federal Aviation Administration's Recommended Exposure Limits for Air-Carrier Crewmembers
Effective Dose Limit
1 mSv for duration of pregnancy and 0.5 mSv in any one month
100 mSv per 5 years and no more than 50 mSv in any one year
(FAA) recommended exposure limits are shown in Table 3 (FAA, 2008). Pregnant crewmembers are most likely to exceed recommended limits. Galactic cosmic radiation is enough to exceed the limits during pregnancy over the course of a few transcontinental round trips.
Neutron Monitor Uses
Once it reaches the atmosphere, a cosmic ray, whether of solar or galactic origin, creates a shower of secondary particles. The shower may reach all the way through the atmosphere to the ground if the initiating particle has enough energy. For purposes of calculating dose rates at altitudes all the way down to sea level, good information about the incoming particle spectrum is required and neutron monitor data currently play a vital role in unfolding the spectrum. As an example, in CARI-7, the FAA’s latest GCR flight dose calculation software, neutron monitor data are used in two places. The data are used both for Forbush modulation, and also within the GCR models to drive solar modulation. Because of the well-known inverse relation between solar activity and GCR flux, using NM data to estimate solar modulation is common practice for GCR model developers (e.g., O'Brien et al., 2003; O'Neill, Golge, and Slaba, 2015). The BO14 GCR models add satellite data and sunspot numbers as additional sources of solar modulation data, while the ISO model is driven by Wolf sunspot numbers and does not capture short-term variations in local GCR flux (O'Neill, Golge, and Slaba, 2015; ISO, 2004). For GCR spectral variation over the course of a day or more a single 18-tube monitor can be used to drive a model with enough accuracy to match in-flight measurements very well. While the HEPAD instruments on NOAA's GOES series satellites could also provide data on these time scales with reasonable accuracy, to capture shorter term variations, such as are needed for studies comparing hourly-averaged instrument measurements to theoretical calculations, only NMs can currently provide the needed short-term accuracy. Multiple monitors/more tubes are useful to improve counting statistics and reliability and are needed to gain detailed spectral information about short term variations such as Forbush decreases and solar particle events (SPEs).
During an SPE, the cosmic ray spectrum incident at any given location in the atmosphere is more difficult to model well. This is because the particle spectrum is constantly changing significantly (even on timescales as short as 1 minute), and because isotropy of the incident flux is often a poor assumption for the first few hours of an event. Data from many monitors, with many tubes each, is needed to maximize accuracy of event assessments. This is particularly true at the start of events, when high-energy particle fluxes are usually highest. For SPEs, a world grid of neutron monitors, in both hemispheres (N and S), is needed to provide a really good picture. Monitors at different altitudes at the same location are also useful. These data from Inuvik, Goose Bay, and Deep River monitors show the problem (Figure 1). Deep River is farthest south and has a slightly higher geomagnetic cutoff, so it could be argued that it's reduced count rate increase relative to that at Inuvik is the result of the softness of the SPE spectrum, but both Goose Bay and Inuvik are near-sea-level stations sited close enough to the magnetic pole that (after correction for local conditions) they would respond essentially identically if event flux were isotropic. Monitors at the same geomagnetic latitude but different longitudes respond differently when SPE flux is highly anisotropic. Using data from multiple high-latitude near-sea-level monitors, along with lower latitude monitors and higher altitude monitors provides information on the shape of the SPE particle spectrum with rigidities above 1 GV, a part of the solar cosmic ray spectrum that satellites do not measure well. Count rates from GOES satellite instruments provide a picture of the low to intermediate energy proton flux, but very limited information the high energy flux spectrum, only enough to make an educated guess at the shape of the spectrum.
Figure 1. Relative increase in count rate relative to GCR background during the SPE of 24-25 May, 1990, at three northern-latitude neutron monitors (Bartol Research Institute, 2015).
With regards to Galactic Cosmic Radiation
Neutron monitors are used for observing solar modulation in real time, including Forbush decreases.
High resolution (hourly) data covers practically the entire jet age of civilian flight (starting in the late 1950s).
Useful for both long term monitoring (solar cycle) and short term variations (Forbush effects).
One monitor with many tubes is enough for basic GCR modelling, if statistics are good and magnetic drift of the cutoff is accounted for, but more monitors add more insight, and may be needed to get adequate statistics if monitors have only a few tubes each.
With regards to Solar Cosmic Radiation
NMs provide the best data for reconstructing SPE particle spectrum anisotropy, which is needed to drive the most accurate and sophisticated SPE flight dose models.
To reconstruct global anisotropic particle flux spectra requires multiple NM at different altitudes and geomagnetic latitudes, both N and S.
NMs are the only data source providing insight into multi-GeV proton and alpha spectra during a SPE; satellite instruments simply do not have enough shielding to discriminate well.
Bartol Research Institute (2015). http://neutronm.bartol.udel.edu/listen/main.html#tell, accessed 10 Oct 2015.
Bureau of Labor Statistics (2016). Occupational Outlook Handbook. Available at: www.bls.gov/ooh/, accessed 11 May 2016.
Copeland, K; Sauer, HH; Duke, FE; Friedberg, W. Cosmic radiation exposure of aircraft occupants on simulated high-latitude flights during solar proton events from 1January 1986 through 1 January 2008. Advan Space Res, 2008, 42(6), 1008-1029.
Dwyer JR, et al. Estimation of the fluence of high-energy electron bursts produced by thunderclouds and the resulting radiation doses received in aircraft. J Geophys Res, 2010; 115: D09206, doi:10.1029/2009JD012039.
Federal Aviation Administration. Order 3900.19B, Chapter 14, Part 1406, Paragraph ‘a’. Washington, DC: Department of Transportation, Federal Aviation Administration, 26 August 2008.
International Commission on Radiological Protection. The 2007 Recommendations of the International Commission on Radiological Protection, Report No. 103.London: Elsevier, 2007.
ISO (International Standards Organization). Space environment (natural and artificial) --Galactic cosmic ray model. ISO 15930:2004. Geneva, Switzerland: ISO; 2004. Available from: www.iso.org/iso/home/store/catalogue_tc/ catalogue_detail.htm?csnumber=37095&commid=46614, accessed 12 June 2013.
National Council on Radiation Protection and Measurements. Ionizing Exposure of the Population of the United States, NCRP Report No. 160. Bethesda, MD, 2009.
O’Brien, K., Smart, D.F., Shea, M.A., Felsberger, E., Schrewe, U., Friedberg, W., Copeland, K. World-wide radiation dosage calculations for air crew members. Advan. Space Res. 31(4), 835-840, 2003.
O'Niel, P.W., Golge, S., and Slaba, T.C. Badhwar - O'Neill 2014 Galactic Cosmic Ray Flux Model Description, NASA/TP-2015-218569. Houston, TX: Johnson Space Center, 2015.
UNSCEAR (United Nations Scientific Committee on the Effects of Atomic Radiation). Sources and Effects of Ionizing Radiation: United Nations Scientific Committee on the Effects of Atomic Radiation: UNSCEAR 2000 Report to the General Assembly, with Scientific Annexes Volume I: Sources. New York, NY: United Nations, 2000.
Warner-Jones, SM; Shaw, KB; Hughes, JS. Survey into the Radiological Impact of the Normal Transport of Radioactive Material by Air. Report NRPB-W39. National Radiation Protection Board (UK); Apr 2003. Available from: www.hpa.org.uk/web/HPAwebFile/HPAweb_C/1194947310807, accessed 29 Jul 2014.
Single-Event Upsets and Microelectronics
(Why neutrons matter to the electronics industry)
Michael Gordon, IBM TJ Watson Research Center
Single-Event Upsets (SEU) are a major reliability issue in modern CMOS devices. They are initiated either by the passage of terrestrial neutrons or by the natural alpha particle radiation from contamination within materials near the transistors, e.g. components of the back end of the line (BEOL) and packaging. Nuclear (spallation) reactions occurring between the terrestrial neutrons and the silicon and other chip materials cause the emission of secondary, highly-ionizing particles. The charge deposited by these secondary particle can cause single, or even multiple, bits to flip. Alpha particles from contamination, and ultra-low energy protons, can also cause SEU’s through direct ionization.
SEU’s in CMOS devices can affect the performance of computer systems (e.g., servers, data clusters, etc.), laptops, and personal electronic devices. Aside from being a nuisance, SEU’s can also have more serious implications as they can also affect medical implanted devices, as well as autonomous vehicles, and passenger airplane avionics.
Semiconductor companies, and end users of high-end chips, often have test campaigns to assess the sensitivity of their devices, circuits, or systems to terrestrial neutrons. The testing methods are usually “accelerated” meaning exposing the devices or systems to an external beam of neutrons (or high energy protons as a proxy for the neutrons), or “life testing”, where typically a computer system, is exposed to the natural terrestrial neutron flux at high elevations, for a period of several months. Subsequently these experiments are repeated underground (with negligible neutron flux) to determine the alpha particle contribution to the SEU rate. The acceleration factors, the ratio of the neutron flux in either the accelerated testing or life testing to the ground-level flux in NY, ranges from ~1E8 to ~ 1.5E1, respectively, as life testing occurs at high elevations.
The chart below, from Autran, et. al., IEEE Radiation Effects Data Workshop, IEEE, 2014, pp. 1-8, shows the neutron component of the SEU for SRAM devices for various scaling nodes from 130 nm, to 40 nm. The units on the vertical axis are FIT/ Mbit where 1 FIT is one bit flip in 1E9 years. The histograms shown in black are the “real-time” or life test data, the lightly shaded histograms are the results of “accelerated” testing using a neutron beam from Triumf (and the yellow histograms are modeling results). The real-time data came from their experiments performed in the French Alps, at the ASTEP facility, 2552 meters above sea level. This facility has real-time neutron monitors (three 3He detectors) used to measure the terrestrial neutron flux during the test campaign. This allows for the real-time assessment of the acceleration factor (~6), rather than an estimate based on geomagnetic rigidity and altitude. This can be important during periods of active sun where the real-time neutron monitors can measure fluctuations in the terrestrial neutron flux.