The major existing problem for deciphering and using the detailed climate record in the Chinese loess-paleosol sequence is understanding the physical and chemical processes by which climate has controlled magnetic susceptibility. Most workers now conclude that pedogenic processes are responsible for most of the variations in magnetic susceptibility in the Chinese loess-paleosol sequences. Studies published during the first half of the quadrennium [ Zheng et al., 1991; Maher and Thompson, 1991; 1992] showed that the enhancement of magnetic susceptibility in paleosols is associated with ferrimagnetic minerals produced during pedogenesis. Many such pedogenic grains are so small that they lack remanence (permanent magnetization), a condition called superparamagnetic (SP), but many SP grains may nevertheless have high intrinsic magnetic susceptibility. Hence the paleosols, the concentration of SP-sized grains, and the past extent of the summer monsoon are all linked.
The influence of pedogenesis on magnetic properties is now
recognized from three principal approaches: (1) spatial patterns
of magnetic susceptibility variations; (2) differences in the
grain sizes and amounts of ferrimagnetic minerals; and (3)
patterns of magnetic susceptibility and 
Be variations.
Magnetic susceptibility of a given paleosol varies across the loess plateau, showing an increase from west to east. This pattern is matched by increased reddening in the paleosols along with a decrease in the thickness of the loess [ Liu et al., 1993]. In addition, magnetite/maghemite grains in the paleosols are more abundant and finer grained than in loess beds. The differences in grain size and in the amounts of SP minerals in paleosols are so great that they are incompatible with a constant rain of magnetic detritus diluted by high loess deposition [ Zheng et al., 1991; Maher and Thompson, 1991, 1992; Hus and Han, 1992; Banerjee et al., 1993; Fine et al., 1993; Heller et al., 1993; Liu et al., 1993; Rolph et al., 1993; Verosub et al., 1993; Evans and Heller, 1994].
The degree of enrichment of SP grains in paleosols has been documented in several ways. Using hysteresis modeling, Maher and Thompson [1992] found that such grains account for more than 90% of the magnetic susceptibility contrast between loess and paleosol. Banerjee et al. [1993] pointed out the inadequacy of frequency-dependent magnetic susceptibility, which senses only a narrow range of SP grain size, and low-field magnetic susceptibility, which can be compromised by paramagnetic substances, to isolate the magnetic signal from pedogenesis. They described a test to estimate the full SP contribution using the thermal demagnetization of saturation isothermal remanent magnetization acquired at low temperature (15 K) in a zero magnetic field. The technique separates the temperature-dependent loss of remanence of SP grains from that of coarse-grained magnetite, as identified by a sharp disordering transition at 120 K. Applied to two widely separated sites, the technique uncovered higher SP fractions within the paleosols than in related loess horizons and higher SP fractions for all samples (except those from the uppermost paleosol) at the more humid site. The relative contributions of SP grains to total saturation remanence at 15 K ranges from as much as 75% in the paleosols to about 20% in loess from the drier site.
Magnetic properties of fractions of sediment separated according to size further corroborates differences in magnetic particle sizes between paleosol and loess. Zheng et al. [1991] showed that the submicron clay fraction of a paleosol from one site is greatly enriched in SP minerals. Similarly, Fine et al. [1993] measured different magnetic susceptibility in five different size fractions from paleosol and loess. The magnetic susceptibility increased with decreasing grain size.
In a different approach, Verosub et al. [1993] and Fine et al. [1993] concluded that most of the magnetic susceptibility signal in paleosols and loess is due to pedogenesis on the basis of combined mineral magnetic methods and a soil-chemistry extraction procedure, known as citrate-bicarbonate-dithionite (CBD) treatment. CBD extraction dissolves small ferrimagnetic grains, both maghemite and magnetite [ Hunt et al., in press], and it has little measurable effect on the relatively large ferrimagnetic grains that are inherited from the parent soil material (primarily magnetite). Using Mössbauer spectroscopy, Singer et al. [1992] demonstrated the selectivity of the CBD treatment of terrace soils from California in removing pedogenic maghemite, including maghemite precipitated directly and maghemite formed via the low-temperature oxidation of magnetite [e.g., Özdemir et al., 1993]. In these soils the CBD-soluble ferrimagnetic phases exhibited behavior characteristic of SP and single-domain (SD) grains (those smaller than about 0.1 şm), whereas CBD-resistant titanomagnetite exhibited behavior characteristic of larger grains that contain more than one magnetic domain, termed multidomain grains [ Fine et al., 1992]. In the Chinese loess-paleosol sequence, CBD treatment diminished the magnetic susceptibility of samples from both loess (about 60-75%, depending on age) and paleosol (>80%). After treatment, mean magnetic susceptibility values are nearly the same for both loess and paleosol. The decrease in magnetic susceptibility even in the loess was interpreted to indicate pedogenic modification of the loess.
Beer et al. [1993] measured both 
Be and magnetic
susceptibility on samples of Chinese loess and paleosol and
developed a time scale for the magnetic susceptibility variations
by matching the 
Be curve to a globally averaged 
O
time scale from marine sediments [ Imbrie et al., 1984]. In
this study the 
Be variations corresponded closely to magnetic
susceptibility; however, the ratios between soil and loess were
much lower for 
Be (2 to 3) than for magnetic susceptibility
(4 to 6). Beer et al. [1993] interpreted the discrepancy as
evidence for the pedogenic origin of most of the magnetic
susceptibility signal in the paleosols. Assuming that the part
of the magnetic susceptibility due to dust flux is proportional
to the 
Be dust flux, the authors calculated the pedogenic
ferrimagnetic contributions to be about 45% in the upper soil
zone (S0) and about 75% in the paleosol (S1) of the last
interglacial period that corresponds to oxygen isotope stage 5,
approximately 130,000 to 115,000 years ago.
The magnetic, chemical, and isotopic studies summarized
above point to a consistent picture of pedogenic production of
ultrafine ferrimagnetic particles in the Chinese paleosols to
account for enhanced magnetic susceptibility. No single method
is yet calibrated as a direct and fully understood climate proxy,
and each method may give an incomplete reading of pedogenesis.
For example, pedogenesis may create grains larger than SP size.
CBD treatment removes maghemite associated with detrital grains
that may have formed before deposition, and it affects ultrafine,
detrital magnetite if present [ Hunt et al., in press].
Finally, the use of 
Be isotopes has assumed proportionality
between the magnetic and the 
Be dust fluxes, and as yet it
neglects variations in atmospheric 
Be production rates.
Toward the goal of using magnetic susceptibility enhancement to
understand better the climate records (e.g., as an indicator of
paleoprecipitation), a next step would be to combine the
different methods on the same samples.