Evidence for a narrow band-limited signal with period near 18.6-years is found in I26 out of 136 yearly total precipitation records, and in 1531 out of 1668 monthly records in the northeastern United States. From 1840 to near the end of the 19th century, rainfall minima in the wavetrain are highly correlated with tidal maxima of the 18.6-year luni-solar tide, the twelfth largest tidal constituent in Newton's theory; the wavetrain then switched phase by 180" and for most of the 20th century rainfall maxima are correlated with tidal maxima at 1917.5, 1936.1, 1954.7, and 1973.3. This bistable phenomenon of atmospheric science was discovered by Currie (1983) in a study of tree-rings from the Patagonian Andes, and OBrien and Currie (1988) have suggested a dynamical explanation in terms of mathematical physics. In terms of yearly rainfall, the mean percentage amplitude modulation of the wave was near +6% until 1940 after which it began to increase rapidly, reaching 10% in the 1960s and 1970s. These results provide a rational explanation for the severe water shortage crisis that occurred at tidal minimum 1964.0 (Namias, 1966;, and reoccurred 19 years later. In addition, a smaller band-limited term with period 10 to 1 I-years is found in a little more than half of the records.

The ideal aim before the mind of the physicist is to understand the external world of reality. But the means which he uses to attain this end are what are known in physical science as measurements, and these give no direct information about external reality. They are only a register or representation of reactions to physical phenomena. As such they contain no explicit information and have to be interpreted. As Helmholtz said, measurements furnish the physicist with a sign which he must interpret, just as a language expert interprets the text of some prehistoric document that belongs to a culture utterly unknown. The first thing which the language expert assumes-and must assume if his work is to have any practical meaning-is that the document in question contains some reasonable message which has heen stated according to some system of grammatical rules or symbols. In the same way the physicist must assume that the physical universe is governed by some system of laws which can he understood, even though he cannot hold out to himselfthe prospect of being able to understand them in a comprehensive way or to discover their character and manner of operation with anything like a full degree of certitude (authors' italics). Max Planck (1932)ABSTRACT Power spectra of 1203 American rainfall records show evidence for two peaks with mean periods of 18.5 1.8 and 106 & 0.6 years, and statistical tests show that both are significant at confidence levels of 99.9 per cent. These are identified as induced by the luni-solar 18.6-year constituent tide and a 10-1 1-year solar cycle variation in the sun's luminosity of the order of 0.1 per cent. Amplitude and phase of the 18.6 year wavetrains are highly non-stationary, and abrupt 180" phase changes in wave polarity are often observed. Wave maxima/minima are generally in phase or out of phase with epochs (epochs are dates of maximum in tidal forcing which occurred at 1898.9, 1917.5, etc.). Polarity maps show that for epochs 1898.9,1917.5,1936.1 and 1954.7 the major part of the nation's landmass experienced droughty conditions, which reached a maximum in both amplitude and areal extent at 1936.1, the Dust Bowl era. At mid-epoch 1964.0 and epoch 1973.3 so many waves changed phase by 180" that the major part ofthe landmass was wet at the epoch. Analogous phenomena have been found in tree-ring data from several continents, and implications for economic science are briefly discussed.

Maximum entropy spectrum analysis of six fish catch series, seven series giving dates of wine harvest, and forty four height of sea-level records from Europe yields evidence for two peaks with periods 18-19 and 10-11 years, the same terms recently reported in European tree-ring chronologies. For each series the two wavetrains are obtained by bandpass filters, summed, superimposed on the trend-like components of the raw data, and plotted against the raw data. The contribution, on average, of the waves to total variance in the raw data varies from a high of 50 per cent for fish catches to a low of 14 per cent for dates of wine harvest (the mean contribution for height of sea-level is 18 per cent). The peaks are identified as the 18.6-year luni-solar, M,, and 10-1 1-year solar cycle, S,, signals, which also have been reported recently in other climate variables, such as air temperature, air pressure, in rain-gauge and tree-ring data from North and South America, South Africa, Tasmania, and New Zealand; and in sea-surface temperature as well as in American crop yield and other economic data. Further experiments show that with 110 years, or more, of data the Yule-Walker method of spectrum analysis has sufficient resolution to resolve both signals. It is also shown that failure to apply a properly designed digital high-pass filter to climate records prior to spectrum analysis usually leads to failure in detecting the terms.

Maximum entropy spectrum analysis of 120 yearly total precipitation records in the Corn Belt

Using a novel data presentation technique, bistable 180" phase switches in 18.6-year induced precipitation over the northeastern United States of America are re-examined. It is found that bistable switching in 19-year wavetrains occurred principally at epochs 1898.9, 1917.5, and 1954.7, although some occurred at mid-epoch 1908-2, and at epochs 1936.1 and 1973.3 (epochs are dates of maximum in the luni-solar 18.6-year tide). The 180" phase switch at 1898.9 occurred along the northern Atlantic seaboard, and by epoch 1954.7 this invading 'cell' from the east blanketed virtually the entire region. OBrien and Currie (1988) have provided a construct in mathematical physics which can explain how such sudden readjustments in the standing wave pattern for air pressure occur on subcontinental scales, and such adjustments are found in other climatic parameters such as air temperature and pressure. Results for seven tree-ring chr.onologies in the region are also given, and are found to have serious inconsistencies both among themselves and also with respect to instrumental rain-gauge data.KEY WORDS Luni-solar 18.6-year term in rainfall Bistable 180" phase switches in climate Maximum entropy spectrum analysis