PDF Design and Analysis of Experiments Classical and Regression Approaches with SAS 1st Edition Leonard C. Onyiah download
PDF Design and Analysis of Experiments Classical and Regression Approaches with SAS 1st Edition Leonard C. Onyiah download
https://ebookgate.com/product/design-and-analysis-of-experiments-5th-
ed-edition-douglas-c-montgomery/
ebookgate.com
https://ebookgate.com/product/design-and-analysis-of-experiments-
solutions-manual-6th-edition-douglas-c-montgomery/
ebookgate.com
https://ebookgate.com/product/design-and-analysis-of-experiments-
advanced-experimental-design-1st-edition-klaus-hinkelmann/
ebookgate.com
https://ebookgate.com/product/regression-analysis-with-python-luca-
massaron/
ebookgate.com
Design and Analysis of Ecological Experiments 2nd Edition
Samuel M. Scheiner
https://ebookgate.com/product/design-and-analysis-of-ecological-
experiments-2nd-edition-samuel-m-scheiner/
ebookgate.com
https://ebookgate.com/product/design-and-analysis-of-experiments-
volume-1-2nd-edition-klaus-hinkelmann/
ebookgate.com
https://ebookgate.com/product/analysis-of-correlated-data-with-sas-
and-r-third-edition-shoukri/
ebookgate.com
Leonard C. Onyiah
St. Cloud State University
Minnesota, U.S.A.
CRC Press
Taylor & Francis Group
6000 Broken Sound Parkway NW, Suite 300
Boca Raton, FL 33487-2742
This book contains information obtained from authentic and highly regarded sources. Reasonable efforts
have been made to publish reliable data and information, but the author and publisher cannot assume
responsibility for the validity of all materials or the consequences of their use. The authors and publishers
have attempted to trace the copyright holders of all material reproduced in this publication and apologize to
copyright holders if permission to publish in this form has not been obtained. If any copyright material has
not been acknowledged please write and let us know so we may rectify in any future reprint.
Except as permitted under U.S. Copyright Law, no part of this book may be reprinted, reproduced, transmit-
ted, or utilized in any form by any electronic, mechanical, or other means, now known or hereafter invented,
including photocopying, microfilming, and recording, or in any information storage or retrieval system,
without written permission from the publishers.
For permission to photocopy or use material electronically from this work, please access www.copyright.
com (http://www.copyright.com/) or contact the Copyright Clearance Center, Inc. (CCC), 222 Rosewood
Drive, Danvers, MA 01923, 978-750-8400. CCC is a not-for-profit organization that provides licenses and
registration for a variety of users. For organizations that have been granted a photocopy license by the CCC,
a separate system of payment has been arranged.
Trademark Notice: Product or corporate names may be trademarks or registered trademarks, and are used
only for identification and explanation without intent to infringe.
Visit the Taylor & Francis Web site at
http://www.taylorandfrancis.com
Preface xxi
Acknowledgments xxv
Author xxvii
vii
viii Contents
Index 817
Preface
This book is intended for students who have had a mix of 100 and 200 level
classes of statistics, covering exploratory data analysis, basic descriptive data
analysis, basic probability, statistical distributions, confidence intervals, and
hypothesis testing. Knowledge of algebra of matrices, determinants, solutions
of simultaneous linear equations, and quadratic equations will be helpful. This
book contains the required computations. Graduate students in biology, psy-
chology, computer science, the physical sciences (physics and chemistry), and
the different fields of engineering have felt comfortable in my classes of STAT
440/540, which were taught from material in this book.
Books written on design and analysis of experiments have hitherto mostly
emphasized the classical approaches to the modeling and analysis of exper-
imental data. Design and Analysis of Experiments: Classical and Regression
Approaches with SAS not only includes classical experimental design theory
but also emphasizes regression approaches while teaching data analysis using
the invaluable tool, SAS, which is the widely used software in statistical data
analysis.
With the growth of statistical computing and availability of cutting-edge
software, the time has come for a new approach for the design and analysis of
experiments. Consequently, in addition to all the classical manual analyses in
this book, explicit SAS programs have been written to analyze the responses
of each experiment, in addition to presenting outputs that result from
the executions of the programs. This book presents explicit SAS codes for the
analyses of responses in the examples of designs presented in the text. The
SAS codes cover classical analyses and analyses based on regression models.
The example SAS codes presented in this book are aimed at providing help for
students and any user in writing their own SAS programs. The purpose of pre-
senting these codes is to help students/users become versatile in their ability
to analyze data, acquire the skills to do it manually or by using SAS, and writ-
ing their own programs and interpreting the outputs. Exercises provided at
the end of each chapter enable students/users to have hands-on SAS program-
ming experience in analyzing data for both classical and regression approaches.
This book, in addition to teaching design and analysis of experiments, is a
useful tool for learning statistical computing with SAS, which will prepare the
student/user for future work environments in which SAS is the software
used routinely for statistical data analysis. Chapters dealing with classical
designs are followed by the regression equivalents of the chapter. For instance,
xxi
xxii Preface
require two one-semester courses to treat each experimental design in this text
thoroughly.
I have provided a solutions manual for all the exercises in this book. This
manual can be purchased with the book to aid in understanding the material.
Exercises included in the book provide a mix of questions requiring manual
and SAS analyses. The problems cover classical and regression approaches.
Doing the exercises should help to induce a well-rounded understanding of
the theory.
Acknowledgments
xxv
xxvi Acknowledgments
and statistics; Jessica Vakili, project coordinator; and Richard Tressider, the
project editor for their invaluable contributions to the production of this book.
Lastly, but above all, I thank Jesus Christ, my Lord, for providing me with
health, strength, and knowledge to complete the writing and production of
this book.
Author
xxvii
Chapter 1
Introductory Statistical Inference
and Regression Analysis
Population. The entire collection of attributes that are under study is called
a population. The attributes may refer to characteristics of objects such
as pencils, steel rods, cars, etc. They may also refer to characteristics of
human or animal subjects.
Sample. The subset of the population that is used in the actual study is
called a sample.
Statistical inference. The general field of statistical endeavor that uses sam-
ples and sample characteristics to make decisions (make inference) about
populations and population characteristics is called statistical inference.
1
2 Design and Analysis of Experiments
n
for θ that maximizes the likelihood function is i=1 xi /n. The estimator of θ
n
is the statistic θ̂ = i=1 Xi /n. The observed value of θ̂, say in an experiment,
n
i=1 xi /n is an estimate of θ.
Example 1.1
n
Show that s2 = i=1 (yi − ȳ)2 /n is a biased estimator of σ 2 .
Solution:
⎛ n ⎞
(yi − ȳ)2 n
E(s2 ) = E ⎝ i=1 ⎠= 1 E yi2 − E(nȳ 2 )
n n i=1
σ2 σ2 nμ2 + σ 2
Note: E(ȳ) = μ; E(ȳ)2 = ⇒ E(ȳ 2 ) = μ2 + =
n n n
Introductory Statistical Inference and Regression Analysis 3
Similarly,
2
n
n
n
E yi2 −E yi = nσ ⇒ E
2
yi2
i=1 i=1 i=1
n
1
= nσ 2 + n2 μ2 ⇒ E yi2 = σ 2 + nμ2
n i=1
1
n
nμ2 + σ 2
E(s2 ) = E yi2 − E(ȳ 2 ) = σ 2 + nμ2 −
n i=1 n
nσ 2 + nμ2 − nμ2 − σ 2 (n − 1)σ 2
= = = σ 2
n n
n
The conclusion reached above shows that s2 = i=1 (yi − ȳ)2/n is a biased
estimator of σ 2 .
Efficient estimator. An efficient estimator for a parameter is an unbiased
estimator with the minimum variance.
A statistic could be an unbiased estimator for a parameter but may
not be an efficient estimator.
Most efficient estimator. The estimator with the least variance of all possible
estimators of a parameter is termed the most efficient estimator for that
parameter.
It is desirable that an unbiased estimator of a parameter should be the most
efficient estimator.
Example 1.2
Show that for a sample of size n taken from a normal population, the sample
mean ȳ is a more efficient estimator of the population mean μ than the sample
median ȳ¯.
Solution:
There are several unbiased estimators of the population mean for the normal
distribution. Apart from the sample mean, another statistic that is unbiased
for the population mean is the median if sample data are normal.
E(ȳ) = μ (μ is the population mean)
E(ȳ¯) = μ (ȳ¯ is the sample median)
It is known from statistical theory (see Wackerly et al., 2002, p. 417) that
σ2
Var(ȳ) = (ȳ is the sample mean, and σ 2 is the population variance)
n
(1.253)2 σ 2
Var(ȳ¯) = (ȳ¯ is the sample median).
n
4 Design and Analysis of Experiments
confidence limits for θ is θ̂ ± Z0.025 σθ̂ . These limits represent the lower and
upper ends of the CI (θ̂ − Z0.025 σθ̂ , θ̂ + Z0.025 σθ̂ ). The lower limit is obtained
when the value of the standard normal variate Z = −1.96, corresponding to
a probability of 0.025 (half of α). The upper limit is obtained when Z = 1.96
is the value of the standard normal variate corresponding to a probability
of 0.975 (equivalent of 1 − α/2) (see Appendices A1 and A2 for table of the
standard normal distribution).
To obtain the CI for a parameter, we need to address four points (1) the
sampling distribution of its estimator, (2) the values of the variate at extreme
ends of the interval based on the confidence of 100(1 − α)%, (3) the computed
value of the estimate from a sample of the population, and (4) the computed
value of standard error for the estimate, if it is known. These values are
substituted in the formula for 100(1 − α)% CI to obtain the desired limits
for the interval. Because the intervals have limits, it is often said that we are
looking for the 100(1 − α)% confidence limits. This is exactly the same thing
as CIs, for once we know the upper and lower limits, we know the interval.
We can easily set up the 100(1 − α)% confidence limits for the mean of the
population based on the sample mean as
σ
ȳ ± Zα/2 √
n
Example 1.3
The variance of the birth-weights of premature babies born in a large
metropolitan hospital is known to be 0.56 kg. Samples of 190 premature babies
were taken to estimate the population mean birth-weight for premature babies
in the metropolis, and a sample mean of 3.02 was found. Obtain (1) the 95%
and (2) the 99% CIs for mean birth-weights of premature babies born in the
metropolis.
Solution:
Let the birth-weights be Y . We know that by central limit theorem, the sample
2
mean ȳ ∼ N μ, σn :
0.56
ȳ = 3.02; σȳ = = 0.0543
190
1. 95% confidence limits = 100(1 − 0.05)% CI
√
= 3.02 ± Z(0.05/2) (σ/ n)
= 3.02 ± Z(0.025) (0.0543)
= 3.02 ± 0.1064
The 95% CI is (2.9136, 3.1264).
Introductory Statistical Inference and Regression Analysis 7
Example 1.4
The following data represent the scores of 20 students in a general education
statistics course involving hundreds of students. Obtain a 95% CI for the mean
of the marks of students in this course. Assume that the scores in this class
are normally distributed. What is the 90% CI for the mean score?
75 45 67 88 90 78 83 89 74 68 44 73 84 91 77 56 64 55 79 72
Solution:
1452
ȳ = = 72.6;
20
1 14522
2
s = (75 + 45 + · · · + 72 ) −
2 2 2
= 198.7 ⇒ s = 14.095
20 − 1 20
s 14.095
√ = √ = 3.15
n 20
s
100(1 − α)% CI = ȳ ± t(n − 1, α/2) √
n
looking for the CI for the difference in means μ1 − μ2 . If we have a large sample
(n ≥ 30), then we can assume that the difference between the means of the
samples ȳ1 − ȳ2 ∼ N (μ1 − μ2 , (σ12 /n1 ) + (σ22 /n2 )) and thus the 100(1 − α)% CI
for μ1 − μ2 is
(ȳ1 − ȳ2 ) ± Zα/2 σȳ1 −ȳ2
where σȳ1 −ȳ2 = (σ12 /n1 ) + (σ22 /n2 ) if the two samples are truly independent.
Unfortunately, we often do not know the values of the population variances
σ12 and σ22 , but when the sample sizes are large, we can substitute the sample
variances for the population variances, so that
s21 s2
σȳ1 −ȳ2 = + 2
n1 n2
Example 1.5
A survey of the heights of 18-year-old students (260 male and 200 female stu-
dents) in the North Jade County was made. The mean height for the males
was 1.68 m with variance 0.076 m. The mean height for the females was 1.57 m
with variance 0.065 m. Obtain the 94% confidence limits for the difference in
the mean heights of 18-year-old males and females from this county.
Solution:
Since the sample sizes are large, each being greater than 30, we can use large
sample theory. This means that the applicable CI is obtained by the formula
s21 s2
100(1 − α)% CI = (ȳ1 − ȳ2 ) ± Zα/2 + 2
n1 n2
0.076 0.065
⇒ 94% CI = 100(1 − 0.04) CI = (1.68 − 1.57) ± 1.88 +
260 200
= 0.11 ± 0.0467
The 94% CI is (0.0633, 0.1567).
When the sample size is not large, the large sample theory does not apply
and we cannot obtain a CI based on the assumption used above. We have to
resort to small sample theory. The difference between the means has a different
distribution, which is not normal. Suppose that sample Y1 contains n1 and
sample Y2 contains n2 observations, then there are two possibilities. If we have
reason to believe that the two small samples might have come from normal
populations with equal variance, and we know this common variance, we can
establish a CI for the difference between the two means by using the formula
1 1
100(1 − α)% CI = (ȳ1 − ȳ2 ) ± Zα/2 σ 2 +
n1 n2
Introductory Statistical Inference and Regression Analysis 9
However, if the variance is unknown, we can pool the variances of the two
samples to obtain an estimate of the common variance. Then the pooled
variance is estimated by
The standard error for the difference in the two means is obtained as
1 1
σ(ȳ1−ȳ2 ) = sp +
n1 n2
The 100(1 − α)% CI for the difference in the two means is obtained by using
the formula
1 1
(ȳ1 − ȳ2 ) ± t(n1 + n2 − 2, α/2)sp +
n1 n2
The second case arises when we have no reason to believe that the two
populations have equal variance. Here, we are comparing two independent
samples, just like the one considered earlier, but the sample sizes are small
(each of n1 and n2 , is less than 30). We will deal with this case later in this
chapter. To find the CI for the difference between the two means when this
condition arises, refer to the method used in Example 1.16.
When we believe that the variances of the two populations are equal, we
use the formulas presented above. We illustrate with an example.
Example 1.6
The skills of repairmen trained by managers A and B who use different styles
for training repairmen are being compared. A sample of 12 repairmen trained
by manager A was chosen, and the time taken by each repairmen to complete
a task in the repair process was obtained. The times of completion of the same
task by a sample of 14 repairmen trained by manager B were obtained. The
completion times they turned out (in minutes) are as follows:
A: 10, 20, 30, 14, 16, 25, 15, 18, 12, 16, 20, 23
B: 12, 22, 22, 27, 31, 16, 20, 24, 30, 10, 26, 24, 28, 31
10 Design and Analysis of Experiments
Obtain a 99% CI for the difference in the mean performance times for the
repairmen trained by managers A and B.
Solution:
12
12
1 2192
y1i = 219 2
y1i = 4355 s21 = 4355 − = 32.568
i=1 i=1
12 − 1 12
14
14
1 3232
y2i = 323 2
y2i = 8031 s22 = 8031 − = 44.53
i=1 i=1
14 − 1 14
The 100(1 − 0.01)% CI for the difference between the means for the
repairmen trained by managers B and A is obtained using the limits
323 219
− ± t(24, 0.005)σȳ1 −ȳ2 = 4.82 ± 2.797(2.4583) = 4.82 ± 6.88
14 12
were obtained, are large (i.e., n1 , n2 ≥ 30), we can establish CIs for the
difference in the two proportions. The standard error for the difference is
p̂1 (1 − p̂1 ) p̂2 (1 − p̂2 )
σp̂1 −p̂2 = +
n1 n2
Since the sample sizes are large, the corresponding CI for the differ-
ence in the two proportions depends on the normal distribution. Hence, the
100(1 − α)% CI is
Example 1.7
“The country is not yet ready for a female president.” In two surveys of this
political question in 1995 and 1999, the following figures were obtained from
the respondents:
Obtain the 95% CI for (1) the proportion of those who said “no” in 1995
and (2) the difference in proportions of those who said “yes” in 1995 and 1999.
Solution:
To establish the CI, we first calculate the estimates of the proportions from
the samples and then apply the large sample theory since it is relevant in this
situation.
348 0.735729(1 − 0.735729)
1. p̂no = = 0.735729 σp̂no = = 0.020275
473 473
null hypothesis. For instance, we could say that the alternative hypothesis
to the null hypothesis stated above is “there is a difference in the mean
yield of the crop under the two fertilizer treatments.” We usually denote the
alternative hypothesis by H1 or HA.
Type I error. This error occurs when we reject the null hypothesis when it
should have been accepted. In other words, type I error is made when
we fail to accept the null hypothesis when it is true.
Type II error. If we accept the null hypothesis when it should have been
rejected (or when the alternative is true), we make a type II error. A
good statistical test of hypothesis should be designed in such a way
as to minimize the sizes of the two errors. However, in practice, it is
often discovered that as we seek to minimize one error, the other error is
increased. A practical decision is usually reached by determining which of
the errors is more serious and this error is then minimized at the expense
of the less serious error, thereby achieving a compromise.
H0 : μs = μ0 vs H1 : μs = μ0
ys = ȳ
ȳ − μ0
Zc = √
σ/ n
The above test has a two-sided or two-tailed alternative. We can also have
tests with one-sided or one-tailed alternatives. The test statistic obtained for
this test is compared with appropriate tabulated values of Z. This Z has the
standard normal distribution with mean 0 and variance 1. The value read
from the table depends on the chosen level of significance or size of type I
error, α.
Decision rules. There are three versions of possible hypotheses that can
be stated for the above problem (Section 1.1.8), and the three sets of
hypotheses can be tested using the same test statistic Zc . Two one-sided
alternatives are available along with the two-sided alternative stated ear-
lier. The choice of alternative hypotheses determines the differences in
the sets of hypotheses that can be tested. The choice is made according
to the interest of the researcher. Although the same test statistic would
do for the three versions of the test, the decision rules are different for
the three versions of the hypotheses.
Introductory Statistical Inference and Regression Analysis 15
Case 3: Two-sided; H0 : μ = μ0 vs H1 : μ = μ0
The critical or rejection regions and their sizes for the three cases
mentioned above are illustrated in Figures 1.1 through 1.3. We use an
example to illustrate the test of hypothesis under the normal distribution.
Example 1.8
It is known that the resting heart rates of a large number of students after
a given period on a treadmill are normally distributed with a variance of 53.
To test the hypothesis that the mean rate is 77/min, the following random
Acceptance region
(size 1 α) Rejection region
(size α)
Acceptance region
Rejection region (size 1 α)
(size α)
Acceptance region
Rejection region (size 1 α) Rejection region
(size α/2) (size α/2)
FIGURE 1.3: Two-sided test with rejection regions to the right and left.
66 75 92 72 72 92
68 62 87 75 62 76
75 90 79 76 80 60
83 75 92 65 73 65
74 83 68 65 70
88 80 68 66 62
68 73 60 82 91
58 62 53 58 86
Carry out this test against an alternative that the mean rate is less than
77, using a level of significance α = 5%.
Solution:
First, we set up the null hypothesis and the alternative as
H0 : μ = 77 vs H1 : μ < 77
2
We know from statistical theory that ȳ ∼ N μ, σn .
The required test statistic Zc is distributed as standard normal variable
with mean 0 and variance 1 as
ȳ − μ
Zc = √ ∼ N (0,1)
σ/ n
The values of Z associated with a given level of significance α can be read
from the tables of the normal distribution.
Now,
44
1
44
3227
yi = 3227 ⇒ ȳ = yi = = 73.34091
i=1
44 i=1 44
73.34091 − 77
Zc = = −3.33397
(53/44)
Introductory Statistical Inference and Regression Analysis 17
Acceptance region
Rejection region (size 1 0.05 0.95)
(size 0.05)
3.33 1.65
Using α = 5%, we read from the table of N (0,1) that Zα = 1.65, so that
−Zα = −1.65.
Since our test corresponds to Case 2 above, we compare Zc with
−Z0.05 . Clearly, Zc = −3.33397 < −Zα = −1.65, so we reject the null hypoth-
esis and accept the alternative that the mean heart rate is less than
77/min. Alternatively, we can make our decision by drawing the figure of
the normal distribution and identifying the rejection and acceptance regions,
so that if Zc falls into the rejection region, we reject the null hypothesis;
otherwise, we accept it. This is illustrated in Figure 1.4.
Example 1.9
It is thought that a laboratory population of fruit flies is made up of flies of
about equal number of gray and black hues. A random sample of 300 flies
yielded 163 black flies. Can we conclude that the proportion of flies are equal
for the two colors? Use levels of significance (1) 5% and (2) 1%.
Solution:
To set up null and alternative hypotheses, we know that number of black
flies follows the binomial distribution. For large n, the number of trials,
the outcome is approximately normally distributed with mean np and vari-
ance npq if p = probability of a black fly and q = probability of a gray fly.
Since np = (1/2)(300) = 150 and npq = (1/2)(1/2)(300) = 75, the resulting
hypothesis is
H0 : p = 0.50 vs H1 : p = 0.50
18 Design and Analysis of Experiments
Acceptance region
Rejection region (size 1 0.05 0.95) Rejection region
(size 0.05/2 0.025) (size 0.05/2 0.025)
1.96 1.96
The number of outcomes of the experiment is used to obtain the test statistic
Zc as follows:
y − np 163 − 150
Zc = √ = √ = 1.50
npq 75
From the tables of the standard normal distribution (Appendices A1
and A2) for α = 5%, Zα/2 = 1.96 and −Zα/2 = −1.96. By the decision rule,
Case 3, we see that Zc lies between −1.96 and +1.96, so we conclude that the
null hypothesis H0 should not be rejected. This means that the proportions
of black and gray flies are probably equal. The decision rule applied here is
depicted in Figure 1.5.
Example 1.10
A biotech company claims that the time it takes for a unit biomass to degrade
is normally distributed with mean of 1500 h and variance of (110 h)2 . This is
disputed by a rival company, which believes that a unit biomass degrades in
less time. The lifetimes of a random sample of 20 units of these biomasses
were studied, and their mean degrading times was found to be 1494 h. Can
we support the claim of the producer at 5% level of significance?
Solution:
The null and alternative hypotheses to be tested are
H0 : μ = 1500 vs H1 : μ < 1500
The test statistic is
ȳ − μ 1494 − 1500
Zc = √ = = −0.2439
σ/ n 110
√
20
From the tables (Appendices A1 and A2), Z0.05 = −1.645, so we do not
reject the null hypothesis. The decision rule applied in this case is represented
in Figure 1.6.
Introductory Statistical Inference and Regression Analysis 19
Acceptance region
Rejection region (size 1 0.05 0.95)
(size 0.05)
1.645 0.24
Case 1: H0 : μ = μ0 vs H1 : μ > μ0
or
20 Design and Analysis of Experiments
Case 2: H0 : μ = μ0 vs H1 : μ < μ0
or
Case 3: H0 : μ = μ0 vs H1 : μ = μ0
we use
ȳ − μ0
tc = √
s/ n
n
where s2 = i=1 (yi − ȳ)2 /n − 1.
Under H0 , tc is t distributed with n − 1 degrees of freedom. The above
statistic will be used for testing the 2 one-sided and the 1 two-sided tests
stated above in Cases 1–3.
Decision rule. Let α% be the level of significance for the tests. Then, here
are the decision rules for the tests:
Case 1: If tc > t(n − 1, α) reject null hypothesis, accept the alternative.
Case 2: If tc < −t(n − 1, α) reject null hypothesis, accept the alternative.
Case 3: If tc > t(n − 1, α/2) or tc < −t(n − 1, α/2) reject null hypothesis,
accept the alternative.
Example 1.11
A manager of a market garden chain was planning for inventory purposes and
believed that a particular shop in the chain sold less than 60 plants per day
on average. He took a random sample of plant sales for 21 days and obtained
the following figures:
71 50 76
62 43 22
63 66 93
68 63 10
81 25 61
61 66 32
59 20 67
Assuming that a daily sale of plants is normally distributed, test the claim
of the manager using a level of significance of 5%.
Solution:
We state the hypothesis to be tested as follows:
H0 : μ = 60 vs H1 : μ < 60
Now, we calculate all the statistics required for the test ȳ, s2 , and tc as
follows:
21
1159
ȳ = yi = 1159 ⇒ ȳ = = 55.19948
i=1
21
Random documents with unrelated
content Scribd suggests to you:
(Poesías, CXLVII)
XXVII
Véase página 242, nota 2.
(Poesías, CLII)
(Poesías, CXLIX)
......................................
(Poesías, CL)
(Poesías, CLV)
Di morte certo, ma non già dell’ora[516]...
(Poesías, CLVII).
XXVIII
Véase página 246.
(Poesías, LXXXI)
NOTAS:
[509] Yo que os he sido dado solamente por una hora, me he dado para siempre
a la muerte. Mientras más ha encantado mi belleza, más lágrimas ha causado:
hubiera sido mejor no haber nacido.
[510] Si alguna vez viví, sólo tú lo sabes, piedra que aquí me guardas. Y si alguno
se acuerda de mí, le parecerá soñar; tan rápida es la muerte, que al que ha sido,
le parece como si nunca hubiera sido.
[511] El que me llora muerto, espera en vano que al bañar mis huesos y mi
tumba, refloreceré como árbol seco y sin frutos; el hombre muerto no renace en
primavera.
[512] Libre de un pesado e importuno despojo, oh mi querido Señor, y
desprendido del mundo, como una barca frágil vuelvo a ti, cansado de la horrible
tempestad a la dulce calma...
[513] Día por día, desde mis primeros años, Señor, fuiste mi guiador y mi auxilio...
[514] Las quimeras mundanas me robaron el tiempo, que se me había dado para
contemplar a Dios...
Mi querido Señor, redúceme a la mitad el camino que sube al cielo, hazme odiar
todo lo que vale en el mundo, y todas sus bellezas a las cuales honro y sirvo, para
ganar con la muerte la vida eterna.
[515] Cargado de años y lleno de pecados...
[516] Seguro de la muerte, pero no de su hora...
[517] Alusión al mal de piedra del cual sufría. “Tre pietre nelle vesica”, según la
explicación de Frey.
BIBLIOGRAFÍA
A.—POESÍAS.
Rime di Michelagnolo Buonarroti, raccolte da Michelagnolo suo
nipote, Giunti, Florencia, 1623.
(Primera edición—defectuosa—del conjunto de las poesías de Miguel
Ángel, hecha por su sobrino nieto Miguel Ángel el joven).
Le Rime di M. A. B. cavate dagli autografi, e pubblicate da Cesare
Guasti, Florencia, 1863.
(Primera edición de las poesías que tiene carácter verdaderamente
histórico).
Die Dichtungen des Michelagniolo Buonarroti, herausgegeben und
mit kritischem Apparate versehen von Dr. Carl Frey, Professor der
neueren Kunstgeschichte an der Universitaet Berlin—mit einer
Portraetradierung von Albert Krüger, und einer Heliographie nach
Francesco da Hollanda,—G. Grote’sche Verlagsbuchhandlung, Berlín,
1897.
(Edición modelo, única exacta y completa, con un admirable
comentario filológico e histórico, una selección de poesías dirigidas a
Miguel Ángel, un cuadro cronológico, extractos de cartas relativas a
las poesías y un índice alfabético).
B.—CARTAS.
Le Lettere di Michel Angelo Buonarroti, pubblicate col Ricordi ed i
Contratti artistici per cura di Gaetano Milanesi, Le Monnier, Florencia,
1875.
A.—DOCUMENTOS CONTEMPORÁNEOS.
Giorgio Vasari. Vite degli architetti, pittori e scultori, 1550 (primera
edición); 1568 (segunda edición).
Ascanio Condivi. Vita di Michel Angelo Buonarroti, Antonio Blado,
Roma, 1553.
Francisco da Hollanda. Cuatro conversaciones sobre la Pintura,
tenidas en Roma, en 1538-1539, arregladas en 1548 y publicadas
por Joaquim de Vasconcellos, traducción francesa en las Artes en
Portugal, por el conde A. Raczynski, Renouard, París, 1846.
Donato Giannotti. Dialoghi de’ giorni che Dante consumò nel cercare
l’Inferno e 'l Purgatorio, compuestos en 1545. Primera edición, 1859,
Florencia.
Paolo Giovio. Michaelis Angeli Vita, publicada primero por Tiraboschi;
Storia della lett. Ital., tomo IX, 1781, Módena.
Benvenuto Cellini. La Vita, escrita entre 1559 y 1562. Primera
edición, 1728, Nápoles.
Benedetto Varchi. Due Lezzioni, Florencia, 1549.
Benedetto Varchi. Orazione funerale recitata nelle esequie di Michel
Angelo Buonarroti, Giunti, Florencia, 1564.
Francesco Berni. Opere burlesche, Giunti, Florencia, 1548.
Á
Los Corresponsales de Miguel Ángel: I. Sebastiano del Piombo, texto
italiano publicado por primera vez por Gaetano Milanesi, con
traducción francesa de A. le Pileur, librería de El Arte, París, 1890.
Sammlung ausgewaehlter Biographien Vasaris, herausgegeben von
Carl Frey, Tomo II. Le Vite di M. A. B. (Edición crítica de todas las
biografías de Miguel Ángel, compuestas por sus contemporáneos).
Giovanni Gaye. Carteggio inedito d’artisti dei secoli XIV, XV, XVI,
Florencia, 1840.
Daelli. Carte Michelangiolesche inedite, Milán, 1865.
Sammlung ausgewaehlter Briefe an M. A. B., herausgegeben von
Cari Frey, Berlín, 1899.
B.—OBRAS MODERNAS.
Richard Duppa. The Life and literary works of M. A. B., Londres,
1806, 1807.
Quatremère de Quincy. Historia de la vida y las obras de M. A. B.,
París, 1835.
Hermann Grimm. Das Leben Michelangelos; primera edición, 1860,
Hanover, séptima y última, 1900 (con ilustraciones).
Aurelio Gotti. Vita di M. A. B., Florencia, 1875.
La obra y la vida de Miguel Ángel, dibujante, escultor, pintor,
arquitecto y poeta, por Charles Blanc, E. Guillaume, Paul Mantz,
Charles Garnier, Méziéres, A. de Montaiglon, G. Duplessis y Louis
Gonse, París, Gazette des Beaux-Arts, 1876.
C. Heath Wilson. Life and works of M. B., Londres, 1876.
Anton Springer. Raffael und Michelangelo, 1878, Leipzig.
Ludwig von Scheffler. Michelangelo, eine Renaissance Studie, 1892,
Altenburg.
John Addington Symonds. The Sonnets of M. A. B. and T.
Campanella, Londres, 1878.
John Addington Symonds. The Life of M. A. B., Londres, 1893.
Carl Just. Michelangelo, 1900, Leipzig.
Corrado Ricci. Michelangelo, 1901, Florencia.
Ernst Steinmann. Die Sixtinische Kapelle, 1905, Bruckmann, Munich,
tomo II (para la iconografía de Miguel Ángel y de Vittoria Colonna).
Dr. Paul Garnault. Los retratos de Miguel Ángel, 1913, París
(Fontemoing).
Henry Thode. Michelangelo und das Ende der Renaissance, tomo I.
Grote, Berlín, 1902; tomo II, ibid, 1903. (Esta obra considerable,
todavía no terminada, es el ensayo más importante que se haya
hecho de un estudio psicológico de Miguel Ángel y de su tiempo. Es
de lamentarse en esta obra, además de una obsesión wagneriana
desagradable y un poco exagerada, el abuso de las categorías
abstractas y de las divisiones escolásticas, que obscurecen el tema
en lugar de aclararlo, y que aumentan el desorden de la
composición, demasiado compacta. La he aprovechado con
abundancia, así como las admirables ediciones y estudios de Carl
Frey).
LA HISTORIA DE MI INFANCIA
La Historia de mi Infancia fué comenzada en el Otoño de 1851, en
Tiflis, y concluida en Piatigorsk, en el Cáucaso, el 2 de julio de 1852.
Es curioso observar que en el cuadro de esta naturaleza que lo
embriagaba, en plena vida nueva y en medio de los peligros
inquietantes de la guerra, ocupado en descubrir un mundo de
caracteres y de pasiones que le eran completamente desconocidos,
Tolstoi se haya vuelto hacia los recuerdos de su vida pasada en esta
primera obra. Pero cuando escribió Infancia se encontraba enfermo,
su actividad militar bruscamente interrumpida; y durante los
prolongados ocios de la convalecencia, dolorido y solo, estaba en
una disposición sentimental de espíritu en la cual, ante sus ojos
enternecidos, revivía lo pasado[555]. Después de la agotadora
tensión de los últimos años, tan desagradables, era para él dulce de
reanimar “el período maravilloso, inocente, poético y alegre” de la
edad primera, y rehacerse un “corazón de niño, bueno, sensible y
capaz de amor”. Por otra parte, con el ardor de la juventud y sus
ilimitados proyectos, dado el carácter cíclico de su imaginación
poética, que raramente concebía un tema aislado y para la cual las
grandes novelas no eran sino eslabones de una larga cadena
histórica, fragmentos de vastos conjuntos que no pudo nunca
ejecutar[556], Tolstoi no podía ver en las narraciones de Infancia, en
aquel momento, sino los primeros capítulos de una Historia de
cuatro épocas, que también comprendería su vida en el Cáucaso y
concluiría, sin duda, en la revelación de Dios por la Naturaleza.
Más tarde, Tolstoi fué muy severo para las narraciones de Infancia, a
las cuales debió una gran parte de su popularidad.
“¡Es esto tan malo,—decía a Birukov;—está escrito con tan poca
honestidad literaria!... De eso no se puede sacar nada”.
En esta opinión estuvo solo. La obra manuscrita, enviada sin nombre
de autor a la gran revista rusa Sovremennik (El Contemporáneo), fué
en el acto publicada (el 6 de septiembre de 1832) y tuvo un éxito
unánime que después han confirmado todos los públicos de Europa;
y sin embargo, no obstante su encanto poético, su finura de
colorido, su emoción delicada, es fácil de comprender que más tarde
haya desagradado a Tolstoi. Le desagradó por las mismas razones
que gustaba a los demás; porque es necesario decirlo claramente: a
excepción de la pintura de algunos tipos locales y en un pequeño
número de páginas, que sorprenden por el sentimiento religioso o
por el realismo en la emoción[557], la personalidad de Tolstoi se
acusa débilmente en esta obra. Se extiende por sus páginas un
dulce, un tierno sentimentalismo, que después siempre le fué
antipático y que proscribió de sus demás novelas. Lo reconocemos, y
reconocemos este “humor” y estas lágrimas, que vienen de Dickens.
Entre sus lecturas favoritas, de los catorce a los veintiún años,
Tolstoi señala en su Diario: “Dickens, David Copperfield. Influencia
considerable”. Todavía en el Cáucaso releyó este libro.
Otras dos influencias señala él mismo: Sterne y Toepffer. “Entonces
estaba yo bajo la inspiración de ellos”[558].
¿Quién habría pensado que las “Nouvelles Genevoises” fueron el
primer modelo del autor de La Guerra y la Paz? Y basta, sin
embargo, saberlo para descubrir en las narraciones de Infancia la
bonhomía afectuosa y zumbona de Toepffer, trasplantada a una
naturaleza más aristocrática.
Encontró Tolstoi, al principiar, que ya era conocido, y su personalidad
no tardó mucho en afirmarse. Adolescencia (1853), menos pura y
menos perfecta que Infancia, descubre una psicología más original,
un sentimiento de la naturaleza más vivo y un alma atormentada,
alma con la cual Dickens y Toepffer se habrían sentido a disgusto. En
La Mañana de un Señor (octubre de 1852)[559], el carácter de Tolstoi
aparece netamente formado, con la intrépida sinceridad de sus
observaciones y su fe en el amor. Entre los notables retratos de
campesinos que pinta en esta novela se encuentra ya el bosquejo de
una de las más hermosas visiones de sus Cuentos Populares: el
anciano en el colmenar,[560] aquel viejecito bajo el abedul, con las
manos extendidas, los ojos en alto, su cabeza calva luciente al sol, y
en torno de ella, las abejas doradas que revuelan sin picarle,
formándole una corona...
LOS COSACOS
Y por encima de estas obras se levanta, cumbre la más alta de esa
primera cadena de montañas, una de las más bellas novelas líricas
que Tolstoi haya escrito, el canto de su juventud, el poema del
Cáucaso, Los Cosacos[564]. El esplendor de las nevadas montañas
que destacan sus nobles líneas sobre el cielo luminoso, llena con su
música el libro entero. Y la obra es única por esta flor del genio, “el
todopoderoso dios de la juventud, como dice Tolstoi: ese ímpetu que
ya no se recobra más”. ¡Cuál torrente primaveral!... ¡Qué efusiones
de amor!
“¡Yo amo, amo tanto!... ¡Bravo! ¡Bueno!... repetía y deseaba llorar.
¿Por qué? ¿quién era bravo? ¿qué amaba? No lo sabía bien”[565].
Esta embriaguez del corazón se derrama desordenadamente. El
héroe Olenine, que ha llegado como Tolstoi a sumergirse en el
Cáucaso en una vida de aventuras, y que se ha enamorado de una
joven cosaca, se abandona al torbellino de sus aspiraciones
contradictorias. Ora piensa que “la felicidad está en vivir para los
otros, en sacrificarse” ora que el “sacrificio de sí mismo no es más
que una tontería”; entonces no está lejos de creer con el viejo
cosaco Erochka, que “todo está permitido y que Dios ha hecho todo
para placer del hombre. Nada es pecado. Gozar con una hermosa
muchacha no es pecado, es la salud”. Pero, ¿qué necesidad tiene de
pensar? Le basta con vivir. La vida es todo bien, toda felicidad, la
vida todopoderosa, la vida universal: la Vida es Dios. Un naturalismo
ardoroso enciende y devora al alma. Perdido en el bosque, en medio
de “la vegetación salvaje, de la multitud de bestias y de aves, de
nubes de moscos, entre la sombría verdura, en el aire cálido y
perfumado, entre pequeños caños de agua turbia que espejean por
doquiera bajo el follaje”, a dos pasos de las emboscadas del
enemigo, Olenine “es embargado de pronto por un sentimiento tal
de felicidad sin causa alguna, que, fiel a una costumbre de su
infancia, se persigna y se pone a dar gracias a alguien”. Como un
fakir indio goza al confesarse que está solo y perdido en este
torbellino de vida que le envuelve, en el cual miríadas de seres
invisibles acechan en este momento su muerte, ocultos por todas
partes, en que millares de insectos zumban en torno suyo, se
llaman:
“¡Por aquí, por aquí, compañeros! ¡Aquí hay alguien a quien picar!”
Y era bien claro para él que allí ya no era un gentilhombre ruso, de
la sociedad de Moscú, amigo y pariente de éstos y aquéllos, sino
simplemente un ser cualquiera como el mosquito, el faisán, el ciervo;
como aquéllos que vivían, que rondaban en torno suyo.
—Como ellos, yo viviré y moriré. Y la yerba crecerá encima de mí...
Y su corazón se llena de alegría.
Vive Tolstoi, en esta hora de juventud, en un delirio de fuerza y de
amor a la vida. Se abraza a la Naturaleza y se funde en ella; en ella
vierte, adormece y exalta sus penas, sus alegrías y sus amores[566];
mas nunca esta embriaguez romántica afecta a la lucidez de su
mirada. En ninguna otra página como en este ardiente poema los
paisajes son pintados con tamaño vigor, ni los tipos con más verdad.
La oposición entre la naturaleza y el mundo, que informa el fondo
del libro y que será toda su vida uno de los temas favoritos en las
ideas de Tolstoi, un artículo de su Credo, le hace encontrar ya, para
fustigar la comedia del mundo, algunos de los ásperos acentos de la
Sonata a Kreutzer[567]. Pero no es menos verídico con relación a
quienes ama, y los seres de la naturaleza, la hermosa cosaca y sus
amigos, son contemplados en plena luz, con sus egoísmos, sus
avaricias, sus engaños, con todos sus vicios.
Una ocasión iba a presentársele para poner a prueba esta veracidad
heroica.
Our website is not just a platform for buying books, but a bridge
connecting readers to the timeless values of culture and wisdom. With
an elegant, user-friendly interface and an intelligent search system,
we are committed to providing a quick and convenient shopping
experience. Additionally, our special promotions and home delivery
services ensure that you save time and fully enjoy the joy of reading.
ebookgate.com