Use the Miller method to test whether or not the VARIANCES of
  two samples are the same (see Section 5.2 in text).
Miller's idea is to apply a `stratified' jackknife to the
  log sample variance.
Compare Miller's P-value and 95% confidence interval with a
  stratified bootstrap P-value and 95% confidence interval.

Diffusion of tritiated water (that is, water containing H^3)
  across human chorioamnion as a test of placental permeability
The first sample is at term (approximately 39 weeks)
  The second sample is at 12-26 weeks.
Table 4.1 (p110) in Hollander & Wolfe, 2nd 3dn

At term (approx 39 weeks) (Xs, m=10)
(Xbar=1.313, SampStdev=0.441, Sampvar=0.195)
  0.80  0.83  1.89  1.04  1.45  1.38  1.91  1.64  0.73  1.46

At 12-26 weeks (Ys, n=5)
(Ybar=0.976, SampStdev=0.197, Sampvar=0.039)
  1.15  0.88  0.90  0.74  1.21

Set gamma = SampVar(X)/SampVar(Y)
Observed gamma = 0.194668/0.03893 = 5.00046

JACKKNIFE PROCEDURE:
First, get the jackknife pseudovalues for the log sample variance
  for each sample:
LogSampVar of Sample 1 (n=10) for full sample: -1.63646
Pseudovalues of Sample 1 (LogSampVar, n=10):
  -1.0531  -1.2555  -0.5620  -2.2607  -2.5887
  -2.6708  -0.3905  -2.0645  -0.5115  -2.5723
LogSampVar of Sample 2 (n=5) for full sample: -3.24599
Pseudovalues of Sample 2 (LogSampVar, n=5):
  -3.2830  -4.0893  -4.2068  -2.0265  -2.0807

JACKKNIFE TEST of H_0:gamma=1  for  gamma=Var(X)/Var(Y):
Sample means and sample variances of the two samples of
  pseudovalues of log sample variances:
  Abar=-1.593  Bbar=-3.137    S2Avar=0.874  S2Bvar=1.105
  Unpooled variance of Abar-Bbar=0.308  Unpooled StDev=0.555
V1=S2Avar/m (etc) and the Miller two-sample Z-statistic:
  V1=0.087  V2=0.221    Z=2.781   P=0.0054 (2-sided)

For gamma(X,Y)=s2(X)/s2(Y):
Bias-corrected gamma estimator and symmetric 95% CI for gamma:
  4.6847   (1.5773, 13.9135)   Observed gamma=5.0005

Starting random-number seed: 123456

BOOTSTRAP PROCEDURE (n=10,000 bootstrap replications):
Use a `stratified bootstrap' to generate an array
  of gamma(star) bootstrap resampled values for
  gamma(star) = s(X)^2(star)/s(Y)^2(star)
Filling the bootstrap ``gamma(star)'' array:
Sorting the bootstrap ``gamma(star)'' array:
The first 10 gamma(star) values are:
   0.052   0.351   0.418   0.441   0.544
   0.551   0.572   0.587   0.617   0.645
The last 10 gamma(star) values are:
   2518.736   2546.194   2575.056   2627.792   2680.569
   2708.833   2818.069   2822.903   2831.125   3008.403
Bootstrap median bias (Obs_Gamma=5.00046):  4069/10000 = 0.407

For gamma(X,Y)=s2(X)/s2(Y):
Bootstrap median and bootstrap 95% CI for gamma:
  5.6289   (2.0966, 48.7321)   Observed gamma=5.0005

Bootstrap P-value for H_0:gamma=gamma0=1  (gamma(observed)=5.00046):
(Twice the proportion of resampled gammas <= 1.)
  P = 2*21/10,000 = 0.0042 (2-sided)

Random numbers used:  150,105
Starting random-number seed: 123456