Figure Captions.

Figure 1. Quaternary faults, primarily from England and Molnar [1997a,
and references therein]. EHS, Eastern Himalayan Syntaxis; LC, Langcang
fault; LMS, Longmen Shan; QJ, Qujiang Fault; WSG, Weihe Shanxi Graben
system; RR, Red River fault. Circles provide locations of earthquakes
mentioned in text, where 1 is 1970 Tonghai M_{w}=7.26 event, which
occurred in the vicinity of the Qujiang Fault, 2 is 1988 Langcang-Gengma
(M_{s}=7.6) series, and 3 is M_{w}=7.6 Manyi (Tibet) earthquake.

Figure 2a. Strain rates inferred from Kostrov summation of Quaternary
fault slip rates (open principal strain rate axes), and corresponding
spatial averages of predicted strain rates (solid principal strain
rate axes) given by bicubic Bessel interpolation of fitted velocity
values on curvilinear grid. Fitted strain rate field is a self-consistent
estimate determined in a leastsquares inversion in which both strain
rates and GPS velocities are matched by model strain rates and velocity
fields, respectively. Amurian B, Amurian Block; Ordos B, Ordos Block;
Sunda B, Sunda Block; WSG, Weihe Shanxi Graben system. 

Figure 2b. Model velocity field relative to Eurasia obtained by bicubic
Bessel interpolation of model rotation vector function values \mathbf{W}(\widehat{\mathbf{x}})
at the nodes in the grid in Figure 2a. Both GPS velocities and Quaternary
strain rates are matched in the inversion procedure (minimization
of equation (8)). Error ellipses are 95% confidence. WSG, Weihe Shanxi
Graben.

Figure 3a. Fit of model velocity field (thin solid vectors), in a model
Eurasian reference frame, to GPS vectors in Himalaya [Larson et al.,
1999] (thin open vectors), at Lhasa, and other Tibetan GPS stations.
IGS station observation at Lhasa is the thick open vector. Thick shaded
vectors for GPS observations at Lhasa location, Wenquan, and Shiquanhe
are from Zhu et al. [2000]. The frames of reference determined in
the inversion that rotated these GPS observations into this model
Eurasian reference frame are given in Table 1. All error ellipses
for both model and observation are for 95% confidence.

Figure 3b. Fit of model velocity, in model Eurasian reference frame
(thin solid vector), to GPS velocity at station IISC (Bangalore).
Model velocity field is from the solution presented in Figure 2. IGS
velocity is the thick open vector, and the thick shaded vector (directly
beneath the IGS vector) is from Zhu et al. [2000]. The NUVEL-1A India
Eurasia vector predicted at this site is shown as the longer shaded
vector (with no error ellipse shown). The frames of reference determined
in the inversion that rotated these GPS observations into this model
Eurasian reference frame are given in Table 1. All error ellipses
are for 95% confidence interval. 

Figure 3c. Fit of model velocity field (thin solid vectors), in model
Eurasian reference frame, to GPS velocities. Model velocity field
is from the solution presented in Figure 2. The thick open vectors
are from IGS, the thick shaded vectors are from Zhu et al. [2000],
the thin shaded vectors (with no arrow head) are from Yu et al., [1999],
and the thin open vectors (with large arrow head) are from Heki et
al. [1999]. The frames of reference determined in the inversion that
rotated these GPS observations into this model Eurasian reference
frame are given in Table 1. All ellipses are for 95% confidence ellipse.

Figure 3d. Fit of model velocity field (thin solid vectors), in model
Eurasian reference frame, to GPS velocities in SE Tibet, SW China,
and northern Indo-China. Thick open vectors are IGS velocities, the
thick shaded vectors are from Zhu et al. [2000], darker shaded vectors
(with no arrow head) are from Simons et al. [1999], and thin open
vectors are from King et al. [1997]. The frames of reference determined
in the inversion that rotated these GPS observations into this model
Eurasian reference frame are given in Table 1. All ellipses are for
95% confidence ellipse.

Figure 3e. Fit of model velocity field (thin solid vectors), in model
Eurasian reference frame, to GPS velocities in Tien Shan and Tarim
Basin. Thin white vectors are from Abdrakhmatov et al. [1996], thick
white vector is IGS velocity, and thick shaded vectors are from Zhu
et al. [2000]. The IGS velocity at POL2 is very small (\sim 1 mm/yr)
and does not have a thick arrow. The frames of reference determined
in the inversion that rotated these GPS observations into this model
Eurasian reference frame are given in Table 1. All error ellipses
are for 95% confidence.

Figure 4a. Velocity field solution presented in Figure 2 but defined
relative to India. Solution shows convergence vectors everywhere normal
to the Himalaya. The solution also shows fanning of velocities within
Tibet, reflecting distributed extension and shear there, and rotation
of crustal material around the Eastern Himalayan Syntaxis. Ellipses
represent 95% confidence.

Figure 4b. Model fit to GPS vectors of Larson et al. [1999] and other
Tibetan stations shown in Figure 3a, except placed into Indian frame
of reference. Ellipses represent 95% confidence. 

Figure 5a. Velocity field difference between solution obtained from
Quaternary strain rates alone and solution obtained from the joint
fitting of Quaternary strain rates and GPS vectors. The error ellipses
of the difference vectors are from the velocity field obtained from
Quaternary alone and are for 1\sigma . 

Figure 6a. Contours of twice the dilatational strain rate (2\dot{\sigma })
from model solution presented in Figure 2. Values of 2\dot{\sigma }
are plotted because they are equal to the negative of the vertical
strain rate, assuming zero volume change. Light shaded areas represent
positive dilatation corresponding to crustal thinning or regions of
dominant extension. All darker shaded regions represent areas of negative
dilatation corresponding to zones of compression or crustal thickening.
First contour line on either side of zero (dashed line) is \pm 1\times 10^{-16}
s^{-1} and these contours are also dashed. Contours beyond this are
multiples of 2\times 10^{-16} s^{-1} and are dotted. For negative
dilatation rates above 10^{-15} s^{-1}, the contour interval is 10^{-15}
s^{-1}. RR, Red River Fault, WSG, Weihe Shanxi Graben system.

Figure 6b. Contoured values for shear strain rates associated with
the strike-slip component of faulting from model solution presented
in Figure 2. Contour interval is 2\times 10^{-16} s^{-1}. See text
for explanation. 

Figure 6c. Rotation rates relative to Eurasia, defined by equation
(6). Negative values are clockwise (dark shaded regions) and positive
values are counterclockwise (light shaded regions). Contour intervals
are the same as in Figure 6a. For reference, 1\times 10^{-15} rad
s^{-1} is equivalent to 1.8^{o} Myr^{-1}.

Figure 7a. Moderate and large-sized earthquake focal mechanisms, mostly
since 1963. Large focal mechanisms are for M_{w}>7.0. Some of these
large events (M_{w}>7.0) are historic events (pre-Wolrld-Wide Standardized
Seismograph Network) from Molnar and Deng [1984] and Molnar [1992],
but all other mechanisms are post-1963 from Molnar and Chen [1983],
Baranowski et al. [1984], Ni and Barazangi [1984], Nelson et al. [1987],
Chen and Nabelek [1988]; Molnar and Lyon-Caen [1989], Ekström and
England [1989], Holt et al. [1991]; Fan et al. [1994], Bernard et
al. [2000], and the centroid moment tensor (CMT) catalog [Dziewonski
et al., 1981, 1996] for most events after 1977.

Figure 7b. Areas listed in Tables 4a-4c for predicted scalar moment
calculation from solution in Figure 2. AB, Amurian block; AT, Altyn
Tagh; CC, central China; CH, central Himalaya; CTS, central Tien Shan;
CY, south Xianshuihe and central Yunnan; EB, eastern Baikal; EC, eastern
China; EHS, Eastern Himalayan Syntaxis; EST, Eastern Syntaxis, Tibet;
ETS, eastern Tien Shan; G, Grabens; GP, Gobi Platform; H, Haiyuan;
HKP, Hindu Kush-Pamir; IB, Indo-Burma Ranges; KL, Kun Lun; KP, Kazakh
Platform; M, Mongolia; MP, Makran-Pakistan; NC, northern China; NP,
northern Pamir; NS, Nan Shan; NT, northern Tibet; OB, Ordos Block;
QB, Qaidam Basin; QSDS, Qinling Shan and Dabie Shan; SCC, south China
coast; SCS, south China Sea; SP, Shan Plateau; SRR, southern Red River
Fault; S, Saigang Fault; SCT, south central Tibet; SY, southern Yunnan
and eastern Myanmar; TB, Tarim Basin; WB, Western Baikal; WH, western
Himalaya; WIC, western Indochina; WSG, Weihe Shanxi Grabens; WTK,
western Tibet and Karakorum; WTS, western Tien Shan; XHE, Xianshuihe.