Window functions and their applications in signal processing 1st edition by Prabhu ISBN 9814463086 978-9814463089instant download
Window functions and their applications in signal processing 1st edition by Prabhu ISBN 9814463086 978-9814463089instant download
https://ebookball.com/product/window-functions-and-their-
applications-in-signal-processing-1st-edition-by-prabhu-
isbn-9814463086-978-9814463089-20280/
Digital Signal Processing and Applications With the C6713 and C6416
DSK 1st Edition Rulph Chassaing ISBN 9780471690078 978-0471690078
https://ebookball.com/product/digital-signal-processing-and-
applications-with-the-c6713-and-c6416-dsk-1st-edition-rulph-
chassaing-isbn-9780471690078-978-0471690078-12010/
https://ebookball.com/product/digital-signal-processing-
principles-algorithms-and-applications-3rd-edition-by-john-
proakis-dimitris-manolakis-0133942899-978-0133942897-17128/
https://ebookball.com/product/digital-signal-processing-
handbook-1st-editon-by-vijay-madisetti-
isbn-0849385725-978-0849385728-17724/
https://ebookball.com/product/digital-signal-processing-signals-
systems-and-filters-1st-edition-by-andreas-antoniou-
isbn-0071454241-978-0071454247-13614/
Conceptual Wavelets in Digital Signal Processing 1st Edition by Lee
Fugal ISBN 0982199457 9780982199459
https://ebookball.com/product/conceptual-wavelets-in-digital-
signal-processing-1st-edition-by-lee-fugal-
isbn-0982199457-9780982199459-11562/
https://ebookball.com/product/digital-signal-processing-2nd-
edition-paulos-r-diniz-eduardoa-b-da-silva-sergiol-netto-
isbn-0521887755-978-0521887755-11086/
https://ebookball.com/product/handbook-of-formulas-and-tables-
for-signal-processing-1st-edition-by-alexander-poularikas-
isbn-0367400316-978-0367400316-17690/
https://ebookball.com/product/digital-signal-processing-
laboratory-2nd-edition-by-b-preetham-kumar-
isbn-1439817375-978-1439817377-13384/
https://ebookball.com/product/signal-processing-noise-1st-
edition-by-vyacheslav-tuzlukov-0849310253-9780849310256-17552/
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.
The Open Access version of this book, available at www.taylorfrancis.com, has been made available under a
Creative Commons Attribution-Non Commercial-No Derivatives 4.0 license.
Trademark Notice: Product or corporate names may be trademarks or registered trademarks, and are used
only for identification and explanation without intent to infringe.
Prabhu, K. M. M.
Window functions and their applications in signal processing / K. M. M. Prabhu.
pages cm
Includes bibliographical references and index.
ISBN 978-1-4665-1583-3 (hbk. : alk. paper)
1. Signal processing--Digital techniques. I. Title.
TK5102.9.P73 2014
621.382’2--dc23 2013031807
. oreword . . . . . . .
F . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiii
.Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xv
.Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xix
.Abbreviations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxi
vii
viii Contents
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 365
Foreword
xiii
xiv Foreword
xv
xvi Preface
When Dr. Gagandeep Singh of Taylor & Francis Group/CRC Press asked me
to write a book for them, I agreed to consider his request. Since my time was
limited, I said that I could write a monograph on Window Functions and Their
Applications in Signal Processing, instead of a textbook, on the condition that he
should get the opinion on this topic from a few experts in the signal processing
area. I provided two sample chapters and the proposed table of contents of
the monograph. These details were sent to six professors for their feedback
with a detailed questionnaire. To my surprise, all the feedback (except one),
were quite positive and this enthused me to write this monograph.
I profusely thank Professor Fred Harris, San Diego State University,
USA, Professor Jonathon Chambers, Loughborough University, UK, Pro-
fessor Palaniappan Ramaswamy, University of Essex, UK, Professor Wai
Lok Woo, Newcastle University, UK, and an anonymous reviewer, for
giving me encouragement and suggestions/criticisms on the proposed
monograph.
I thank the authorities of the Indian Institute of Technology (IIT) Madras,
India, for granting me one semester sabbatical during July–November 2011,
as well as providing me with all the necessary facilities in bringing this mono-
graph to its present form. I profusely thank Prof. J.W.M. Bergmans, Chairman
of the Signal Processing Systems group, Technical University Eindhoven
(TU/e), Netherlands, for providing me travel grant to visit TU/e at different
times.
I wish to extend my thanks to dr. ir. A.W.M. van den Enden, who was
earlier with Philips Research and Philips Semiconductors, Eindhoven, The
Netherlands (as a lead signal processing expert). He is now an independent
consultant where he works closely for Philips Research, NXP, Catena Radio
Design, SRON, and others. He tirelessly and meticulously read through all the
chapters and we had detailed discussions on each chapter when I visited him
at TU/e, Eindhoven last year. He also introduced me to the topic of canonic
signed digit (CSD) representation. I express my whole-hearted appreciation
to him for spending his valuable time overlooking his own personal work.
I wish to thank my research students who were an integral part of this
project. They are Basil Abraham, S. Abhijith Varma, H.N. Bharath, Sunit
Sivasankaran, P. Vikram Kumar, Sanjay Kumar Sindhi, and T. Srinivas Rao,
all from the Department of Electrical Engineering, IIT Madras, India. I also
thank my summer intern, Swathi Nagaraj from M.I.T., Manipal, India. All of
them diligently worked toward achieving the goal to my satisfaction. I pro-
fusely thank all of them for their dedication toward perfection. I also thank my
long-time friend, philosopher and guide, Prof. Umapathi Reddy, presently
xix
xx Acknowledgments
xxi
xxii Abbreviations
We note that the unit impulse function can be related to the unit step
signal as:
du(t)
δ(t) = . (1.3)
dt
3. Complex exponential signal: This is represented by the following
function:
x(t) = ceat , (1.4)
u(t)
t
0
FIGURE 1.1
Unit step signal.
Fourier Analysis Techniques for Signal Processing 3
(a) δ (t)
t
0
(b) δΔ (t)
1/Δ
t
0 Δ
FIGURE 1.2
(a) Unit impulse signal. (b) Unit delta function.
3. Complex exponential:
x[n] = cα n (1.8)
T
x1 [n] ←→ y1 [n] = T(x1 [n]) (1.11)
T
x2 [n] ←→ y2 [n] = T(x2 [n]) (1.12)
T
ax1 [n] + bx2 [n] ←→ y3 [n] = T(ax1 [n] + bx2 [n]) (1.13)
and if y3 [n] = ay1 [n] + by2 [n]. Here, a and b are any arbitrary con-
stants.
FIGURE 1.3
System.
Fourier Analysis Techniques for Signal Processing 5
T
x[n] ←→ y[n] (1.14)
T
x[n − n0 ] ←→ y1 [n]. (1.15)
and
y[n] ≤ By ∀n, 0 ≤ By < ∞. (1.17)
∞
This section has provided a brief overview of the signals and systems needed
to understand the following chapters. In case the reader is interested in more
details, refer to Refs. [1–4].
Equations (1.24) and (1.25) together are known as the continuous-time Fourier
transform (CTFT) pair.
We can construct the FT of a periodic signal directly from its Fourier series
representation. The resulting transform, given below, consists of a train of
impulses in the frequency-domain.
∞
F
→ e−jt0 X(j).
x(t − t0 ) ← (1.28)
8 Window Functions and Their Applications in Signal Processing
If the scaling factor has a magnitude greater than unity, then the signal
is compressed in the time-domain, while its frequency spectrum gets
expanded. For |a| < 1, exactly the converse happens, that is, the time-
domain signal is expanded and the spectrum is scaled down.
Fourier Analysis Techniques for Signal Processing 9
7. Parseval’s theorem: This gives the relation between the energy (E) of
a signal in the time-domain and the frequency-domain. We can use
this property to easily compute the energy of a signal by integrating
the squared magnitude of its FT.
∞
1 ∞ 2
E= |x(t)|2 dt = X(j) d. (1.36)
−∞ 2π −∞
(a) x(t)
t
0
(b) X( jΩ )
1/a
Ω
0
(c) arg[X( jΩ)]
π/2
−π/2
FIGURE 1.4
FT pair of e−at u(t). (a) Time-domain signal. (b) Frequency-domain magnitude plot. (c) Frequency-
domain phase plot.
Figure 1.4 shows the signal x(t), the magnitude, and phase responses
of the CTFT of e−at u(t).
2. Determine the CTFT of a two-sided exponential signal given by:
x(t) = e−a|t| , a > 0.
∞
X(j) = e−a|t| e−jt dt
−∞
0 ∞
= eat e−jt dt + e−at e−jt dt
−∞ 0
0 ∞
= e(a−j)t dt + e−(a+j)t dt
−∞ 0
1 1
= +
a − j a + j
2a
= 2 .
a + 2
Fourier Analysis Techniques for Signal Processing 11
(a) x(t)
t
0
(b) X( jΩ )
2/a
1/a
Ω
–a 0 a
FIGURE 1.5
FT pair of e−a|t| . (a) Time-domain. (b) Frequency-domain.
The signal x(t) and the magnitude response of the FT of e−a|t| are
depicted in Figure 1.5. The phase is zero, since the signal is even
symmetric in the time-domain.
3. Find the Fourier transform of the sinusoidal signals:
a. x(t) = cos(0 t)
b. x(t) = sin(0 t)
Note that signals such as sinusoids that exist for all time are not
absolutely integrable. Absolute integrability is a sufficient condition
for the existence of FT, but it is not a necessary condition. These
difficulties can be solved by introducing Dirac delta function in the
frequency-domain as detailed below.
a. x(t) = cos(0 t)
1 j0 t 1 1
x(t) = e + e−j0 t = ej0 t + e−j0 t .
2 2 2
∞
= π δ( − 0 ) + πδ( + 0 ).
1 j0 t 1 1
x(t) = e − e−j0 t = ej0 t − e−j0 t .
2j 2j 2j
1 1
a1 = , a−1 = − for k = ±1 and ak = 0, for k = ±1. (1.39)
2j 2j
Therefore,
∞
π π
= δ( − 0 ) − δ( + 0 ).
j j
∞
ejτ/2 − e−jτ/2
X(j) = x(t)e−jt dt =
−∞ j
sin(τ/2) τ
=2 = τ sinc .
2π
Here, we have used the fact that sinc(t) = sin(πt)/(πt). The time-
domain and frequency-domain plots are shown in Figure 1.7.
Fourier Analysis Techniques for Signal Processing 13
(a) X ( jΩ )
Ω
–Ω o 0 Ωo
(b) X(jΩ )
π
j
Ω
–Ω o 0 Ωo
–π
j
FIGURE 1.6
CTFT of sinusoidal signals. (a) CTFT of cosine signal. (b) CTFT of sine signal.
1, 0 ≤ t ≤ τ
x(t) =
0, otherwise.
∞ τ
e−jτ/2 jτ/2
X(j) = x(t)e−jt dt = 1.e−jt dt = e − e−jτ/2
−∞ 0 j
sin(τ/2)
= 2e−jτ/2
τ
= τ e−jτ/2 sinc .
2π
Here also, we have used the fact that sinc(t) = sin(πt)/(πt). By com-
paring the above result with the result of the previous example, it can
be observed that there is only a phase shift involved in the FT. The
same result can also be obtained by using the time-shifting property
of the CTFT.
14 Window Functions and Their Applications in Signal Processing
(a)
1
0.9
0.8
0.7
0.6
x(t)
0.5
0.4
0.3
0.2
0.1
0 –τ τ
0
2 t 2
(b)
τ
X( jΩ )
2π/τ
0
–6π/τ–4π/τ –2π/τ 4π/τ
6π/τ
0
Ω
FIGURE 1.7
FT of a rectangular pulse. (a) Time-domain. (b) Frequency-domain.
1, || ≤ W
X(j) =
0, || > W.
W
1 1 2 sin(Wt)
x(t) = e d =
jt
2π −W 2π t
Fourier Analysis Techniques for Signal Processing 15
(a) 1.5
0.5
Amplitude
−0.5
−1
−1.5
−1 0 1 2 3 4 5
t (s)
(b) 1.5
0.5
Amplitude
–0.5
−1
−1.5
−1 0 1 2 3 4 5
t (s)
FIGURE 1.8
Figure for Example 7. (a) Input x(t). (b) dx(t)/dt.
sin(Wt)
=
πt
W Wt
= sinc .
π π
dx(t)
= [δ(t) − δ(t − 1) − δ(t − 3) + δ(t − 4)] . (1.40)
dt
The impulse train is shown in Figure 1.8(b). Now taking the CTFT
on either side of the above relation, and applying the differentiation
property, we obtain
jX(j) = 1 − e−j − e−j3 + e−j4
= e−j2 ej2 − ej − e−j + e−j2
= e−j2 ej2 + e−j2 − (ej + e−j ) .
2
x(t) = .
t2 + 1
F 2a
e−a|t| ←
→ .
a2 + 2
2 F
→ 2π e−|| .
←
t2 + 1
TABLE 1.1
Basic CTFT Pairs
Signal Fourier Transform
∞ ∞
k=−∞ ak e jk0 t
2π k=−∞ ak δ( − k0 )
ej0 t 2π δ( − 0 )
cos(0 t) π [δ( − 0 ) + δ( + 0 )]
sin(0 t) π
j
[δ( − 0 ) − δ( + 0 )]
x(t) = 1 2π δ()
δ(t) 1
δ(t − t0 ) e−jt0
∞ 2π ∞
n=−∞ δ(t − nT) k=−∞ δ( − 0 )
⎧ τ
T
⎪
⎨1, |t| ≤
2 τ
x(t) = τ sinc
⎪
⎩0, |t| > τ 2π
2
⎧
⎨1, 0 ≤ t ≤ τ
τ
x(t) = τ e−jτ/2 sinc
⎩0, otherwise 2π
⎧
⎨1,
W Wt || ≤ W
sinc X(j) =
π π ⎩0, || > W
1
u(t) + π δ()
j
1
e−at u(t) , Re{a} > 0
a + j
1
te−at u(t) , Re{a} > 0
(a + j)2
tn−1 −at 1
e u(t) , Re{a} > 0
(n − 1)! (a + j)n
2a
e−a|t| , Re{a} > 0
⎧ a2 + 2
⎪
⎪−1, t<0
⎪
⎨ 2
x(t) = 0, t=0
⎪
⎪ j
⎪
⎩1, t>0
2
2π e−||
t2 + 1
1 2π
ak = x̃[n]e−j N kn . (1.42)
N n=<N>
Here, k = <N> denotes that k can take any N consecutive values. The terms
in the series are harmonics of the fundamental frequency, 2π/N. For the
discrete-time case, the Fourier series is always convergent, since it is only
a finite summation.
We now extend the frequency-domain representation to include more
general aperiodic signals. Let us define a function X(ejω ) as
∞
Comparing Equations 1.42 and 1.43, we can see that ak s are the samples of
X(ejω ), spaced at ω0 = 2π /N in the frequency-domain as
1
ak = X(ejkω0 ). (1.44)
N
As N → ∞, ω0 becomes infinitesimally small. This indicates that X(ejω ) is
sampled with spacing ω0 → 0. Thus, X(ejω ) can be viewed as a continuous
function. Similarly, the summation in Equation 1.41 is carried out over N
consecutive intervals of width ω0 = 2π /N and the total interval of integration
has a width of 2π. Hence, unlike the CTFT where range is over the whole
real axis, the DTFT requires only ω values in the interval [0, 2π ].
The DTFT can be derived by taking the CTFT of a sampled signal. The
IDTFT and the DTFT expressions are given by
π
1
x[n] = X(ejω )ejωn dω . (1.45)
2π −π
∞
Equation 1.45 represents the inverse DTFT (IDTFT), also known as the syn-
thesis formula, while Equation 1.46 gives the DTFT, known as the analysis
formula. The discrete-time periodic signals can be included within the frame-
work of DTFT by interpreting the transform of a periodic signal as an impulse
train in the frequency-domain as
∞
2π
X(e ) =
jω
2π ak δ ω − k − 2πl . (1.47)
l=−∞ k=<N>
N
Fourier Analysis Techniques for Signal Processing 19
The DTFT of x[n] is said to exist only if Equation 1.46 converges in some sense.
There are two types of convergence (also known as summability) which are
defined as follows.
∞
F F
x[n] ←
→ X(ejω ) and y[n] ←
→ Y(ejω ). (1.50)
F
ax[n] + by[n] ←
→ aX(ejω ) + bY(ejω ) . (1.52)
20 Window Functions and Their Applications in Signal Processing
F
→ e−jωn0 X(ejω ) .
x[n − n0 ] ← (1.53)
However, note that both X(ejω ) and e−jωn0 X(ejω ) have the same mag-
nitude responses with only a phase shift introduced due to the
time-shifting property. Therefore, delaying the time sequence has
the effect of shifting the phase of its transformed version, while the
magnitude response remains the same.
4. Frequency shifting property:
F
ejω0 n x[n] ←
→ X(ej(ω−ω0 ) ) . (1.54)
5. Time reversal property: When the sequence x[n] is flipped (i.e., mirror
image over −n), then
F
→ X(e−jω )
x[−n] ← and (1.55)
F
→ X ∗ (ejω ),
x[−n] ← only if x[n] is real. (1.56)
F
x[n] ∗ y[n] ←
→ X(ejω )Y(ejω ). (1.57)
F
→ e−jωnd .
δ[n − nd ] ← (1.58)
For any arbitrary input signal x[n] and the impulse response h[n] of
an LTI system defined as
F
→ H(ejω ) = e−jωnd ,
h[n] = δ[n − nd ] ← (1.59)
F
→ Y(ejω ) = e−jωnd X(ejω ) .
y[n] = x[n] ∗ δ[n − nd ] ← (1.60)
Fourier Analysis Techniques for Signal Processing 21
F d
nx[n] ←
→j X(ejω ) . (1.61)
dω
F
If w[n] ←
→ W(ejω ) (1.62)
and y[n] = x[n].w[n] (1.63)
1 π
then, Y(ejω ) = X(ejθ )W(ej(ω−θ ) )dθ . (1.64)
2π −π
∞ π
1
E= |x[n]|2 = X(ejω )2 dω. (1.65)
n=−∞
2π −π
The generalized Parseval’s relation for two signals x[n] and y[n] is
given by
∞ π
∗ 1
x[k]y [k] = X(ejω )Y ∗ (ejω )dω. (1.66)
k=−∞
2π −π
22 Window Functions and Their Applications in Signal Processing
F
x∗ [n] ←
→ X ∗ (e−jω ). (1.67)
F
x∗ [−n] ←
→ X ∗ (ejω ). (1.68)
x[n] + x [n] F ∗
X(e ) + X ∗ (e−jω )
jω
Real part: Re{x[n]} = ←→ Xe (ejω ) = .
2 2
(1.69)
x[n] − x∗ [n] F X(ejω ) − X ∗ (e−jω )
Imaginary part: jIm{x[n]} = ←→ Xo (ejω ) = .
2 2
(1.70)
x[n] + x∗ [−n] F X(ejω ) + X ∗ (ejω )
Even part: xe [n] = ←→ XR (ejω ) = .
2 2
(1.71)
x[n] − x∗ [−n] F X(ejω ) − X ∗ (ejω )
Odd part: xo [n] = ←→ jXI (ejω ) = .
2 2
(1.72)
(1.74)
Magnitude: X(ejω ) = X(e−jω ) , Phase: X(ejω ) = −X(e−jω ).
(1.75)
1 jω0 n 1 −jω0 n
x[n] = e + e .
2 2
Comparing the above equation with the Fourier series expansion in
Equation (1.22), we observe that a1 = 12 and a−1 = 12 . Thus, the DTFT
of x[n] can be written as a sum of weighted and shifted impulse trains
as follows:
∞
X(e ) =
jω
π [δ(ω − ω0 + 2π l) + δ(ω + ω0 + 2πl)].
l=−∞
Fourier Analysis Techniques for Signal Processing 23
1 jω0 n 1
x[n] = e − e−jω0 n .
2j 2j
∞
π
X(ejω ) = [δ(ω − ω0 + 2π l) − δ(ω + ω0 + 2πl)] .
l=−∞
j
1, |n| ≤ N1
x[n] = .
0, |n| > N1
∞
N1
2N1
1 − e−jω(2N1 +1)
−jωm jωN1
X(e ) =
jω
e e =e jωN1
m=0
1 − e−jω
ejω(N1 + 2 ) − e−jω(N1 + 2 )
1 1
=
ejω/2 − e−jω/2
sin(ω(2N1 + 1)/2)
= .
sin(ω/2)
The time function x[n] and the DTFT are shown in Figure 1.9.
4. Find the DTFT of a shifted rectangular pulse signal defined as
1, 0 ≤ n ≤ 2N1
x[n] =
0, otherwise.
24 Window Functions and Their Applications in Signal Processing
(a) 1
Amplitude 0.8
0.6
0.4
0.2
0
−6 −4 −2 0 2 4 6
Samples (n)
(b)
|X(e jω)| 7
0
−2π −π 0 π 2π
ω (rad)
FIGURE 1.9
Rectangular pulse and its DTFT for N1 = 3. (a) Input x(n). (b) DTFT X(ejω ).
We can write,
2N1
X(ejω ) = e−jωn
n=0
= 1 + e−jω + · · · + e−jω(2N1 )
1 − e−jω(2N1 +1)
=
1 − e−jω
sin(ω(2N1 + 1)/2)
= e−j(ωN1 ) .
sin(ω/2)
⎧
⎪
⎨n + 1, 0 ≤ n ≤ N1 − 1
x[n] = 2N1 − 1 − n, N1 ≤ n ≤ 2N1 − 1
⎪
⎩
0, otherwise.
1, 0 ≤ n ≤ N1 − 1
x1 [n] =
0, otherwise
and x[n] = x1 [n] ∗ x1 [n] ⇔ X(ejω ) = X1 (ejω )X1 (ejω ). (1.76)
sin(ωN1 /2)
X1 (ejω ) = e−jω(N1 −1)/2 . (1.77)
sin(ω/2)
Using the result from Equation 1.77 and substituting into Equa-
tion 1.76, we can obtain X(ejω ) as
2
sin(ωN1 /2)
X(ejω ) = e−jω(N1 −1) .
sin(ω/2)
x(n)
6
0
0 1 2 3 4 5 6 7 8 9 10
Samples (n)
FIGURE 1.10
Figure for Example 5.
26 Window Functions and Their Applications in Signal Processing
6. Consider the signal x[n] = an u[n], |a| < 1. Its DTFT is given by
∞
∞
−jωn
X(e ) =
jω n
a u[n]e = (ae−jω )n
n=−∞ n=0
1
= .
1 − ae−jω
7. Consider the signal x[n] = a|n| , |a| < 1. The DTFT of this signal is
evaluated as follows:
∞
−1
∞
∞
∞
= an ejωn + an e−jωn
n=1 n=0
1 1
= −1+
1 − ae jω 1 − ae−jω
(1 − a2 )
= .
1 − 2a cos(ω) + a2
F d ae−jω
nx[n] ←
→j X(ejω ) =
dω (1 − ae−jω )2
ae−jω 1
Y(ejω ) = +
(1 − ae )
−jω 2 (1 − ae−jω )
1
= .
(1 − ae−jω )2
Fourier Analysis Techniques for Signal Processing 27
Now computing the DTFT for |a| < 1 and applying the result of
Example 6, as well as the time-shifting property, we get
1 aN e−jωN
X(ejω ) = −
1 − ae−jω 1 − ae−jω
1 − aN e−jωN
= .
1 − ae−jω
22. Riseberga.
"Sinä saat lypsää niitä huomenna, jos haluat", sanoi tuo hyvä
rouva tuottavasti. "Auringonkukka lypsää nyt neljättä viikkoa 5
kannua erästään, mutta Metsäruusu on ehtynyt neljään. Ole
varovainen, kun Auringonkukkaa lypsät: sillä on semmoinen paha
tapa, että potkii kumoon kiulun, se ei kärsi ketään muuta kuin
vanhaa Renataa. Ei löydy muuta neuvoa tuollaista vastaan, kuin että
panee märän rievun selkäristimyksen päälle. Ja vielä muistutan, älä
koskaan pidä punaista nauhaa hiuksissasi, kun karjapihaan menet!
Huomaa, miten Turkki sinua katselee! Se on nyt kytkettynä, mutta
mitä nuori härkä punaista kärsisi?"…
Hagar jäi yksin ja istui avoimen akkunansa ääressä, josta oli näkö-
ala puistoon päin. Heinäkuun kuuvalo taisteli katoavan kesäpäivän
viimeisen valonhohteen kanssa, järveä varjosti taivaanrannalta
nouseva pilvi ja heikot syyssalamat välähtelivät. Oli tyyntä, hiljaista
ja lämmintä. Lintujen laulu lehmusten latvoista oli kadonnut;
ainoastaan tuo väsymätön leppälintu antoi vielä kuulla muutaman
yksinäisen liverryksen kevään päiviltä. Se lienee ollut joku uneton
uroslintu; naaras valvoi vartioiden poikastensa yölepoa.
Urban.
23. Regensburg.
"Anna minun sanoa sinulle jotakin!" pyysi tuo lempeä, tuttu, mutta
ei pitkiin aikoihin kuultu ääni. "Tunnetko minun?"
"Et pääse ennen, kuin sinulle olen jotakin sanonut", virkkoi ääni.
"Tahdon sanoa sinulle", jatkoi hän, "että sinä olet suurin narri,
mikä maassa kulkee. Mitä varten sinä suututat parahinta ihmistä,
sinun iso-isääsi, joka on minun isäni setä sekä sinun ja meidän
kaikkien hyväntekiä? Mene oitis hänen luoksensa; pyydä häneltä
anteeksi, ja lupaa, ett'et koskaan tee häntä vastaan! Tiedätkö, mitä
on olla hänen poikanaan? Hän on enemmän kuin kuningas,
enemmän kuin keisari; hän on koko maailman hallitsia; ei yksikään
gojimin ruhtinaista ole niin mahtava kuin hän. Ja sinä tulet yhtä
rikkaaksi ja mahtavaksi kuin hän, jos alistut hänen tahtonsa mukaan.
Mitä sinä juokset kurjien sotiesi ja köyhien sotilaittesi perässä? Etkö
ymmärrä, että kun sinä tulet siksi, kuin iso-isä nyt on, saatat itse
varustaa voimallisimmat sotajoukot, eikä silloin kukaan uskalla sinua
vastustaa? Miksikä et saata totella ja odottaa? Miksikä et tahdo
tunnustaa meidän uskoamme ja olla yksi meistä? Etkö tiedä, että me
olemme valittu kansa ja kaiken maailman valtakunnat tulevat
olemaan meille alammaiset?"
"Sen mukaan, mitä minä tiedän, ei hän vielä ole. Päin vastoin ei
hän edes Ebbastakaan huoli, kuten ennen."
"Rakas sisareni saattaa olla siitä niin varma, kuin että nupusta
puhkeaa kukka. Minä odotan aikaamme, jolloin meitä armollisesti
tahi epä-armollisesti kiitetään palveluksestamme ja saamme mennä."
"Mitä teidän lemmellisyytenne sanoo! Nuori mies! Mutta jos sen
nuoren miehen nimi on herra Kaarle Kustaa?"
ebookball.com