Measuring Constants of Nature Confirmation and Determination in Piezoelectricity

PERGAMON
Studies in History and Philosophy of

Modern Physics 34 (2003) 579-606
Studies in History
and Philosophy
of Modern Physics
Measuring constants of nature: confirmation

and determination in piezoelectricity
Shaul Katzir
Edelstein Center for the History and Philosophy of Science, Technology and Medicine, The Hebrew
University of Jerusalem, Gioat Ram, Jerusalem 91904, Israel
Abstract
Exact measurements are a central practice of modern physics. In certain cases, they are
essential for determining values of coefficients, for confirming theories, and for detecting the
existence of effects. The history of piezoelectricity at the end of the nineteenth century reveals
two different methods of exact measurement: a mathematical versus an "artisanal" approach.
In the former, a scientist first carried out the experiment and later employed mathematical
methods to reduce error. In the latter, a scientist physically manipulated the experimental
apparatus to bypass possible sources of error before its performance. These two approaches
were related to German theoreticians and to French experimentalists, respectively. However,
affiliation with a particular school rather than nationality was the decisive factor in the
differences between the two approaches. Despite differences, adherents of both approaches
sought to attain high precision and to eliminate even small experimental errors.
This history exhibits the complexity and flexibility of experiments and their analysis. It
supports the claim of the "New Experimentalism" that theory does not supply complete
directions for practical experimental decisions. An example for this is found in the way an
experimental error was discovered (after incorrect results had already been published) only by
a comparison to an earlier experiment rather than to a theory.
2003 Elsevier Ltd. All rights reserved.
Keywords: Piezoelectricity; Exact measurement; Confirmation; Experiment; Determination; Jacques
Curie; Pierre Curie; Woldemar Voigt; Eduard Riecke
1. Introduction
During the nineteenth century, exact numerical values and precise results became
increasingly important in physics. By the second half of the century, exact
E-mail address:shaulk@vms.huji.ac.il (S. Katzir).
ARTICLE IN PRESS
580
S. Katzir / Studies in History and Philosophy of Modern Physics 34 (2003) 579606
quantitative measurement using precision instruments had become a distinctive and

essential practice. German physicists called it measuring physics and distinguished
this laboratory activity from that of experimental physics. Measurements were
carried out to obtain precise quantitative data, while experiments did not
necessarily involve quantitative information. Qualitative or approximate quantitative results were usually sufcient for the latter kind of laboratory activities, which
were still the majority. Exactitude was needed in measuring physics, which aimed
not only at accurate results but also at exact values, such as were needed for the
determination of constants of nature. Since precise quantitative values were often
based on mathematical theory, exact measurements were usually carried out by
theoretical or mathematical physicists in Germany. (Jungnickel & McCormmach,
1986, p. 120 and passim; Kuhn, 1977).
One aim of measuring physics was the determination of constants, and such
constants were often considered means to higher ends. Another almost obvious goal
was the conrmation of mathematical theory that could not be tested qualitatively.
Yet, as I show below, these were not the only roles of quantitative experiments.
Moreover, historical understanding of exact measurement at the end of the
nineteenth century requires more than a recognition of their roles. The practice of
such measurements and their relations to other experiments, measurements, and
theories should also be studied. This paper examines the quantitative experiments
done in one new eld of physics, piezoelectricity, in the last two decades of the
nineteenth century. These experiments started with a discovery of quantitative
regularity and continued with determinations of the constants that characterized this
regularity and the conrmation of theoretical predictions. Table 1 presents the major
steps in this process. Focusing on a particular eld will help illuminate the
relationship between measurement and theory among successive experiments that
measured similar quantities. The theory suggested to describe piezoelectricity agreed
well with then-current physics theories. Its conrmation, therefore, did not attract
much attention. Further, it raised neither a controversy nor special experimental
difculties. Its historical examination is a study of a practice of conrmation that
satised the experimentalists and other scientists. Thus, more attention is given here
to the conrmation of the theory accepted by the physical community than to an
earlier conrmation of a theory that was not accepted. It is shown that the limited
experimental testing of the theory did not have any apparent inuence on the
theorys acceptance.
Measuring physics was a German concept, and indeed Germans were most
prominent in the quest for exact numerical results at the time. Nevertheless, even in a
eld dominated by Germans, like piezoelectricity, precise measurements and
determinations were also carried out by French scientists. The French and German
methods, however, were very different, suggesting the existence of disparate
traditions in their approaches to precise experiments and their physical and
mathematical analysis. Their alternative methods of determining piezoelectric
constants demonstrate the differences between the two experimental traditions.
These differences display only one facet of the complexity of quantitative
experimentation at the time. In the following, I show that determining qualitative
ARTICLE IN PRESS
581
Table 1
Major events in the history of piezoelectric measurements
Year
General
1880
1881
Discovery of the phenomenon by the Curies

Empirical rules for the development of
charge by the Curies
Curies determination of quartzs and
tourmalines coefcients
Czermak determination of quartzs
coefcient based on his theoretical account
1887
1890
1891
Measurements
Voigts general theory of piezoelectricity
1894
1898
Riecke and Voigts conrmation and

determination of quartzs and tourmalines
coefcients
Pockels examination of direct electro-optics
and determination of piezoelectric
coefcients
Voigt examination of direct pyroelectricity
results depended on the material apparatus, various levels of theoretical and

experimental analysis, contingency, and, in at least one case, on earlier experimental
results.
2. First determination of the coefcients

The brothers Jacques and Pierre Curie, two young assistants at the Paris Faculty
of Science, discovered piezoelectricity in 1880. Examining a few crystal species, they
found that a variation of pressure applied along asymmetric axes of crystals
developed polar electricity along these axes (Curie, 1880). After initial study of the
crystallographic conditions for the appearance of electricity, they turned to a
quantitative examination of the development of charge by pressure. They placed a
bar of tourmaline between two copper plates, and pressed it along its asymmetric
axis. One plate (A in Fig. 1) was grounded; the other (B) was connected to a sector of
a Thomsons electrometer, whose other sector was connected to a Daniell cell
(D)a battery of known and steady potential. The other pole of the cell was also
grounded. Applying various stresses to tourmaline bars of various dimensions, they
found that the quantity [of electricity] is proportional to the variation of pressure.
Further, For the same variation of pressure the quantity of electricity released is
independent of the dimensions of the tourmaline (Curie, 1881a, pp. 1516). They
soon found that quartz obeyed the same rules and thus assumed that the electric
effect of a variation in pressure was linear in all crystals.
They argued that every crystal had a characteristic coefcient (or coefcients) that
showed the amount of electric charge produced by an increase or decrease in
ARTICLE IN PRESS
582
Fig. 1. From Curie and Curie (1882).
pressure. Soon after convincing themselves of the validity of this relation, the
brothers carried out an experiment to measure the coefcients of tourmaline and
quartz. In establishing the linearity, they measured only the voltage, which they
knew to be proportional to the charge, but not the charge itself. Since they sought
only ratios between pressure and electric charge, that was all they needed. Yet, to
determine the values of the coefcients, they had to nd the quantity of charge,
which is equal to the multiplication of the voltage by the capacity.
The Curies added a cylindrical capacitor (C), connected to the copper plate (B)
that was connected to the electrometer; its other end was grounded (Fig. 1). They
loaded known weights on the crystal1 and replaced capacitors (whose capacities were
calculable) until the charged needle of the electrometer stabilized at its zero point
between the two sectors. At this point, the electric potential in the two sectors was
even, and thus equal to that of the known Daniell cell (1.12 V). Next, they removed
the external capacitor (C) and corresponding weights until the needle stabilized again
at its zero point; i.e., the voltage on the plate was again that of one Daniell. The
difference in the quantity of charge between the two cases was clearly due to the
difference in the weight that pressed the on crystal. Since the voltage was the same,
the difference in charge was a multiplication of the known voltage by the difference
in the capacities, which was the capacity of the cylindrical capacitor (DQ V DC).
Thus, they immediately determined the charge developed per variation of a weight or
force unit (Curie, 1881b; Curie & Curie, 1882).
By determining only differences in quantities, the Curies bypassed complicated
measurements of the capacity of the system. Instead, they needed to know only the
capacity of a known condenser. They used a cylindrical condenser made up of two
[close] pieces [plates], with which one can eliminate the error due to its boundaries
[extr!emit!es], and calculated its capacity from the dimensions by an unspecied
1
Pressures on quartz were applied both in the direction of an (asymmetric) electric axis and
perpendicular to that direction. They found that in a square bar both induce electric charge of the same
quantity (and of an inverse sign) at the ends of a polar axis.
ARTICLE IN PRESS
583
method (Curie & Curie, 1882, p. 250).2 By constructing a null experiment in which
they kept the electrometers needle at zero, the Curies eliminated errors not only in
reading the needles deviation, but more signicantly, in translating its deviation into
units of voltage. Thus, the data analysis in their experiment was simple and did not
require any complicated mathematics. Keeping the voltage low during most parts of
the experiment gave the additional benet of reducing electric leakage. They found
that a pressure of 1 kg generated a charge of 0.062 electrostatic units (statcoulomb)
in quartz and of 0.0531 in tourmaline. Later units were, correspondingly, 6:32
108 ; and 5:41 108 (statcoul/dyne) (Curie, 1881b).3
The publication of these results did not stop the Curies measurement of these
constants during the 1880s.4 Jacques Curie continued measuring the coefcient of
quartz by the same null experiment method at least until the end of the rst decade
of the twentieth century. Marie Curie, Pierres widow, published his results
describing the piezoelectric electrometer that he had designed with his brother in
the early 1880s (and which was used in the study of radioactivity). In this device,
known voltage due to piezoelectricity balanced an unknown electrical source
connected to the other side of a regular electrometer so its dial was left at zero. It
enabled the performance of null deviation experiments with various voltages. This
instrument supplied a motivation for further improved measurements because its
reading was sensitive to the exact value of the piezoelectric coefcient.
In his later measurements, Jacques Curie modied the original experimental
arrangement. He connected the battery and the crystal to the same side of the
electrometer and grounded its other side, keeping its needle at zero. One face of the
quartz (Q in Fig. 2) was grounded, the other (P) connected to the electrometer, a
known cylindrical plate condenser (with a guard ring) (A), and, via a commutator, to
the earth. The condensers other plate (B) was connected to a known battery of ten
Weston cells, invented in 1892 and more reliable than the Daniell cell. While AP
was grounded, a charge q was induced on P through piezoelectricity; the battery
induced a charge q0 on A. These were totally independent of each other. Then a
commutator simultaneously disconnected AP from the earth and connected the
other plate of the condenser (B) to the ground instead of the battery, while the weight
was removed. The electrometers needle would stay at the zero point only if the total
charge on the conductor AP was zero; i.e., when q q0 : The charge q0 is constant and
calculable from the voltage of the battery and the capacity of the condenser, known
from the condensers geometry. By varying the weight (before disconnecting from
2
Unfortunately, the Curies supplied only brief descriptions of their experiments in their published
papers. Archival sources probably cannot help, since no laboratory notebook from that period has been
kept with any of Pierre Curies manuscripts in the various collections examined by his biographer Anna
Hurwic. I thank Ms. Hurwic for a list she made of the contents in the various archives.
3
Apparently the precision of the results was determined by the precision of the calculation of capacity
(including three signicant digits). Thus, the difference in the number of digits between tourmaline and
quartz is arbitrary.
4
In 1882 the Curies presented modied values of 0.063 and 0.053 electrostatic units (1882, pp. 250251).
Yet in 1889, they returned to the earlier value for quartz (they did not mention tourmaline in that
publication, Curie & Curie, 1889, p. 36).
ARTICLE IN PRESS
584
Fig. 2. From the German translation of Curie (1910).
the ground), J. Curie found the weight needed to leave the needle at zero; i.e., the
weight needed to induce a known charge by the piezoelectric effect.5 In 1910, he
determined a value of 6:9 108 ; equal to the current value, from measurements on
several specimens. Later experts attributed the higher value of the later result to
higher attention to the apparatus (improved insulation) and the target specimens
(purity of the specimens and their orientation) rather than to improvements in the
method of measurement (Curie, 1910, pp. 104105; Cady, 1964, pp. 216219).
3. Czermaks determination of the quartz coefcient

After the Curies determination of piezoelectric coefcients, research in the eld
developed in other directions. A converse effect (a linear effect of potential
differences on pressure) was found and studied. The source of the phenomenon and
5
J. Curie probably did not publish the results himself. Marie Curie did not comment on the reasons for
the modication in the experimental design. One advantage of the new design was in eliminating any
oscillation of the needle. In the earlier arrangement, oscillations were inevitable since the electrometer
could not have been connected simultaneously to the cell on one side and the crystal and the condenser on
the other.
ARTICLE IN PRESS
585
its relationship to the phenomenon of pyroelectricitythe generation of electric

polarization due to a temperature changewas another direction of research. Still
another inquiry concerned the relationship between the direction in which pressure
is applied and the distribution of electric charge. Theoretical considerations
and empirical generalizations about the phenomena were mostly qualitative. A
quantitative account of the observed phenomena was limited to the establishment of
its linearity by the Curies. In particular, no mathematical explanation was suggested
for the variation of the effect with the direction of pressure. Paul Czermak suggested
such an explanation in 1887 for quartz, which was the most widely examined crystal
throughout the 1880s. Czermak used the theoretical account to re-determine its
piezoelectric coefcient.
Czermak worked on piezoelectricity while he was a guest researcher at Kundts
physical institute in Strasbourg, probably under Kundts direction (Katzir, 2001, pp.
6465). Unlike former contributors to piezoelectricity, Czermaks expertise was in
theoretical physics. His extraordinary dissertation on molecular paths in
Maxwells kinetic theory of gases, written 2 years before, revealed, according to
his teacher Boltzmann, rare theoretical talent. Still, like most theoretically inclined
physicists of his generation, Czermak also carried out research in the laboratory
(Jungnickel & McCormmach, 1986, vol. 2, p. 68; Poggendorff, 1945, pp. 289290).
Theoretical, or mathematical, physicists in the German sphere were characterized by
their combination of quantitative physical rules with mathematically analyzed
experimental results, and by their interest in exact measurement. Czermaks
piezoelectric research displayed these concerns.
Following the empirical data and the known symmetry of quartz, Czermak
assumed that quartz has three polar axes in the xy plane (perpendicular to the
principal axis); i.e., axes that become polarized by pressure (1,2,3 in Fig. 3). He
adopted the Curies conclusion that the effect was linear. However, unlike the Curies,
who made no assumption about the variation of the effect with orientation, Czermak
tacitly assumed that the effects in the three axes are independent of each other,
though they are not orthogonal. From these assumptions, he obtained an expression
Fig. 3. From Czermak (1887).
ARTICLE IN PRESS
586
for the electric tension (V ) in any direction of pressure in the xy plane as (1887,
pp. 12271229):
p
1
V A 3=4 sin2 60 y sin2 60 y;
where A is a coefcient that depends linearly on the pressure and on an assumed
piezoelectric constant, and (y) on the angle between the direction of pressure and an
adjacent polar axis (Fig. 3).
Czermak tested this expression experimentally. He encircled two quartz cylinders,
whose lengths coincided with the principal axis, with 24 steel strips. He compressed
each cylinder in 48 different directions through its center (i.e., 7.5 each) and
measured the electric voltage in the direction of pressure. He carried out these
measurements with three different weights for each cylinder, thus making a total of
288 measurements. His results presented the then known division of quartz into six
symmetric electric zones of the same distribution of electricity (in absolute values).
He calculated the average voltage induced in the six zones for each direction of
pressure relative to a polar axis, reducing the 288 measurements to 48 values.
Comparing the experimental results with his theory, he found deviations of 2.52
and 2.96 from the directions that he had initially taken to be polar axes by optical
determination. Adjusting the data accordingly, he compared the observed values
with a theoretical calculation based on an average value of the constant A
(calculated in a non-specied manner). Czermak did not offer any systematic
comparison between the calculated and observed values but juxtaposed their
numeric values in a table and charted lines that connected both values on the same
graph. The reader was left to assess the conrmation of the expression. Czermak
himself saw his results as only a rough approximation (1887, pp. 12181232).
After establishing the validity of the theoretical expression (1), Czermak employed
it to redetermine quartzs piezoelectric coefcient in absolute units; i.e., in a standard
system of units independent of any peculiarity of a specic experiment. He designed
a new experiment in which he shaped two square parallelepipeds from the same
quartz specimen. Two parallel surfaces of one bar were cut perpendicular to a polar
axis, and parallel surfaces of the other were cut at an inclination of 15 from the
perpendicular. Applying pressure along these directions, he measured the voltage
between the two pressed surfaces with an electrometer dial whose deviation was
compared before and after each measurement with that due to a known Daniell cell.
He examined the effects caused by three different weights. To eliminate accidental
errors, he repeated his experiment on four different days. These precautions illustrate
his concern with experimental accuracy. While his measurements of the rst prism
(cut perpendicularly to a polar axis) showed the linearity of the phenomenon, those
of the second prism (cut at an inclined angle) did not, and thus failed to yield a
constant. He explained this apparent contradiction as due to a deviation in the
prisms shape (the faces were not exactly parallel). Comparing the results with the
theoretical expressions, he determined the presumed actual directions of these
surfaces and used them in the later calculations. Czermak, thus, used mathematical
analysis to eliminate a systematic error.
ARTICLE IN PRESS
587
Because he compared each reading of the voltage to that of a Daniell cell,

Czermak could rely directly on the electrometers voltage measurement. Yet, the
piezoelectric coefcient related charge, rather than voltage, to weight, so, like the
Curies, Czermak had to determine capacity. Unlike them, he measured the capacity
of the whole system. He used a cylindrical condenser (similar to the one used by the
Curies) whose capacity could be altered at will and was calculable in the various
states from its dimensions by an equation derived by Kirchhoff. Connecting the
electrometer to the condenser (in two states) instead of the apparatus, he obtained a
relationship between the capacities of the latter two and calculated from that the
capacity of the apparatus.6 In this method, he neglected the small capacity of
the electrometer. Instead, he made sure that it would be small. From the capacity, he
determined a value of 0.06142 statcoul/kg (6:27 108 statcoul/dyne) for the
piezoelectric coefcient.7
As Czermak noted, his result agreed well with the value announced by the Curies 6
years earlier. They differed by less than 1% (1887, pp. 12371244). His method,
however, was different from theirs. While Czermaks determination was based on
measurements not only of the target but also of the apparatus and measuring
instrument itself (the electrometer), the Curies based their determination only on two
measurements of the target (replicated to reduce errors). They did not have to
measure the capacity of the system or to translate deviations of the electrometer
needle to units of voltage. Thus, the analysis of their experiment was much simpler.
Czermaks theory of the experimental instrument was not more complicated than the
Curies, but he relied more than the Curies on a theory of the piezoelectric
phenomena itself, since he compared the effects of applying pressure along two
different directions by his (roughly conrmed) theoretical expression (Eq. (1)).
Moreover, he reinterpreted the data (the direction of pressure) according to the same
expression. Thus, the assumption that one constant piezoelectric coefcient exists (in
Eq. (1)) was used to obtain a unique constant and to nd its value. This made his
experiment less reliable. Its reliability was further diminished by a contradiction
between his expression for the electric effect in one direction due to pressure in
another and the empirical data, and later by the rejection of Czermaks theoretical
explanation in favor of Voigts more sophisticated theory (Katzir, 2001, pp. 70, 72,
8284). Probably because Czermaks results were so embedded in his (incorrect)
theoretical calculations, they were no longer used.
4. A general theory of piezoelectricity
Jacques and Pierre Curies determination of the piezoelectric coefcients involved
little piezoelectric theory. They only assumed the linearity of the effect which they
had just established. Czermaks determination of the quartz coefcient involved
6
Czermak did not report his measurement of the relative capacities. Perhaps he compared the deviation
of the electrometer after a constant voltage charged it, with the condenser and with the apparatus.
7
As was quite common at the time, Czermak wrote more digits than the signicant digits in his
experiment, which could hardly yield three signicant digits, let alone four.
ARTICLE IN PRESS
588
more theory. Beyond the linearity, he assumed a specic relationship between the
direction of pressure and the magnitude of the effect. Moreover, in practice, he also
had to rely on his theory in interpreting the data. Later measurements of
piezoelectric coefcients continued to involve piezoelectric theory in their design,
though most did not do so in interpreting the data. Such involvement of theory in the
determination of coefcients should not surprise us. The piezoelectric coefcient
exists and has meaning only in a theoretical framework or at least under an
assumption like that of the Curies about a linear effect. When piezoelectric theory
was elaborated, an assumption like the Curies was not sufcient.
The major subsequent theoretical development was the general piezoelectric
theory formulated by Woldemar Voigt in 1890 (Voigt, 1890). He aimed at a general
mathematical theory based on a few sound assumptions that would embrace all
phenomena of piezo- and pyroelectricity in all crystals. Clearly piezoelectricity
lacked such a theory. Its few theoretical expositions were not comprehensive, lacked
rigor, and did not account well for the observed phenomena. Czermaks
mathematical theory, the only one suggested thus far, was based on an incorrect
treatment of symmetries and suffered from contradictions with empirical data
(Katzir, 2001, pp. 6970).
Voigts was what he and others called a phenomenological theory; i.e., a theory
that accounts for the phenomena by pointing at basic relations between its variables
but does not attempt to explain these relations through any deeper effect or process.
He based his theory of piezo- and pyroelectricity on two hypotheses. First, the
electric effect of strain in every volume element of the crystal is linear, or more
specically that the excited electric moment (dipole moment density, or in short
polarization in modern terminology) is proportional to the deformation (strain).
Second, he assumed that the phenomena are subject to the symmetrical properties of
the crystals in which they occur; i.e., in each crystal, the piezoelectric symmetry is not
of lower order than the crystallographical symmetry. These were general and widely
accepted assumptions. The rst was conrmed experimentally for a few crystals by
the Curies, Hankel, and Czermak. The second is a particular case of a general
principle acknowledged in the physics of crystals, which was sometimes called
Neumanns principle.
A general equation (allgemein Ansatz) of piezoelectricity in any crystal that relates
the three components of electric moment (polarization) a; b and c to the six
components of strain xx in a volume element follows from the theorys rst
assumption:
a e11 xx e12 yy e13 zz e14 yz e15 zx e16 xy ;
b e21 xx e22 yy e23 zz e24 yz e25 zx e26 xy ;
c e31 xx e32 yy e33 zz e34 yz e35 zx e36 xy ;
where the 18 piezoelectric constants ehk are dependent on the character of the
crystals and the location of the coordinate system X ; Y ; Z (Voigt, 1890, pp. 34).8
8
Since the crystal strain tensor is symmetrical (yx xy ), one needs only six (rather than nine)
components of strains in the equation.
ARTICLE IN PRESS
589
Thus, each crystal has up to 18 different constants that relate each component of
polarization to each component of strain. However, by general considerations of
symmetry, according to the theorys second assumption, he reduced this number
considerably for most crystals. For each class of crystals (i.e., a set of crystals that
possesses the same properties of symmetry), he found which constants are identically
zero, which are functions of other constants, and which have independent values.9
These relations are also valid for equations of the polarization due to stress rather
than strain, equations that are more useful in the laboratory. For quartz, the
paradigmatic piezoelectric crystal, for example, Voigt found:
a d11 Xx d11 Yy d14 Yz ;
b 2d11 Xy d14 Xz ;
c 0;
where Xx are the components of stress, and dij were called by Voigt piezoelectric
moduli; the moduli are related to the constants through the crystals elastic
coefcients (pp. 2235). According to this theory, quartz has two independent
piezoelectric coefcients rather than one. The coefcient that the Curies and
Czermak had previously measured corresponds to the moduli d11 : Analyzed in this
way, tourmaline has four coefcients. With similar equations for the various
crystal classes, Voigt succeeded in accounting for all the empirical data and
predicting as yet unobservable phenomena. This theory remains the basis of current
thought.
5. Voigt and Rieckes determination of piezoelectric moduli

Jacques and Pierre Curies announcement of the discovery of piezoelectricity
raised interest in the new phenomena and was followed by experimental work on its
properties. The publication of Voigts theory had a similar inuence, but, this time,
the work was focused on theory. Although scholars might expect experiments to
have a major role in conrming or refuting theories, only a few experiments were
.
made. Moreover, all the empirical tests of the theory were made at Gottingen,
where
Voigt had suggested the general theory, and were personally related to him.
5.1. The confirmation of the theory
In the autumn of 1891, Voigt himself, with his collegue Riecke, performed the only
experiment constructed to test his general theory. The two were, respectively,
.
professors of theoretical and experimental physics at Gottingen.
They undertook not
only the conrmation of the theory but also the determination of the piezoelectric
coefcients of quartz and tourmaline. This effort displayed a condence in the
9
This was the rst application of general considerations of symmetry in piezoelectricity. Previously,
Czermak handled only the special case of quartz, basing his elaboration on the peculiar properties of the
crystal rather than on explicated considerations of symmetry. Invalid assumptions (of the relations
between the effects in the different axes) led him to incorrect results.
ARTICLE IN PRESS
590
validity of the theory because the coefcients make sense only in the context of the
theory.10
In order to conrm the theory, it was necessary to make exact measurements of
crystals, a practice in which Riecke and Voigt had expertise. Riecke had done a few
experiments on pyroelectricity since 1885 (Riecke, 1886). Voigt had performed
experiments on the elasticity of crystals since his student years. Exact determination
of theoretical coefcients in absolute units was a common theme in their works. They
exhibited a high degree of precision in their experiments and mastery of physicalmathematical techniques in analyzing their results. Voigt and Riecke belonged to
two different schools of German physics (Neumann and Weber), yet exact
measurements in absolute units and the elimination of experimental error were
regarded essential in both (Jungnickel & McCormmach, 1986, Vol. 1; Olesko, 1991).
The friendly relationship between Riecke and Voigt (which was unusual between the
experimental and theoretical physicists in the German universities) facilitated their
collaboration.
In their experiment, Riecke and Voigt applied pressure to rectangular crystal
plates. One surface was connected to a Thomsons electrometer (with its other side
grounded) and the other surface to the ground. They measured the deviations of the
electrometer needle due to the electrication of the crystals on an arbitrary scale.
Like Czermak, they compared these readings to the deviation caused by a known
voltage, a Clark cell, which was considered steadier and more reliable than a
Daniell cell (Riecke & Voigt, 1892, pp. 531532). It was less reliable than the
Weston cell that Jacques Curie later used, but that cell was invented only after
they had performed their experiment.
Their apparatus enabled them to load and unload weights and to measure the
effects, assumed to be equal, of both increasing and decreasing the pressure. This
setting enabled them to make many measurements with the same pressure, reducing
the inuence of a single error and improving the accuracy of the experiment.
However, they noticed a deviation in the position of the electrometer needle between
successive measurements. After deciding that the deviation was due to unavoidable
damping and diffusion of electricity, they formulated a mathematical expression that
corrected the reading of the needle. They conrmed it with 70 measurements of both
loading and unloading (1892, pp. 533536). This sequence demonstrated their
meticulous efforts to reach precision by eliminating experimental errors and their use
of mathematical techniques to do so.
Previous quantitative experiments measured the effect of pressure in a direction
that involved only one piezoelectric modulus. The conrmation of the general theory
10
Their condence in the theory was at least partly based on the agreement between its results and the
empirical data collected during the 1880s, i.e., before its formulation. However, Riecke and Voigt wrote,
since the former measurements were conducted with no consideration of a theory and in many cases in
only a qualitative way, they are not suitable for a completely satisfactory examination of the theory. It
therefore remained to perform extended (ausgedehntere) quantitative determinations of several cases that
are accessible (zuganglichen)
to theoretical treatment and after calculating each substances individual
.
piezoelectric constants to compare them with the theoretical laws. They did this on tourmaline and
quartz (1892, p. 524, emphasis added).
ARTICLE IN PRESS
591
required the comparison of the effects of various situations in which more than one
modulus was involved. For quartz, which has two moduli, equal pressures in ve
different directions in the same plane were sufcient. Riecke and Voigt derived a
geometric equation that expressed the measured electric tension in terms of the two
sought coefcients in the ve directions from the piezoelectric equations for quartz
(Eq. (3)). In each direction of pressure, they made three to six measurements and
found the average voltage. From these averages, they used the least-squares method
to calculate the values of the moduli in an arbitrary Clark scale. They used these
values, in turn, to calculate the theoretical predictions for induced electric tension in
their experiment and to compare them with the observed values, concluding that the
agreement is satisfactory(1892, p. 539).11 In showing the agreement of the
expression derived from Voigts theory with the observations, they conrmed
the general theory for the rst time. This was the rst quantitative examination of the
dependence of the electric effect on the direction of the stress.
The measurements on tourmaline were somewhat more complex, since it has four
rather than two piezoelectric coefcients. Applying pressure in different directions in
one plane would not sufce to nd the value of all its coefcients. Riecke and Voigt
therefore measured eight different combinations of directions of pressure and the
faces on which the effect was observed in four crystal bars. This time, they did not
develop a general expression for all the cases but compared each measurement with
its specic theoretical expression. In each measurement, they expressed the potential
in Clark units and found the average for each combination. These averages led to
eight equations with four unknowns. The piezoelectric moduli were derived directly
from this system of equations. In such a derivation, they could not use an error
analysis method like the method of least squares. Instead, they showed that
independent determinations of the coefcients led to similar results12 and concluded
that the agreement between their values can be described as satisfactory.13 Thus,
their observations corroborated the relations deduced from Voigts general theory
and conrmed it. Still, the use of the method of least squares for quartz made the
conrmation in that case more satisfying (even by Riecke and Voigts own
standards) than that of tourmaline (1892, pp. 539544).
5.2. German versus French methods of determination
The measurements in arbitrary Clark scale were sufcient for the conrmation
of piezoelectric theory. Determination of its coefcients in absolute units did not add
to its conrmation except in demonstrating that they are true constants of the
crystals by comparing them with other observations. Nevertheless, Riecke and Voigt
11
Indeed, the maximum deviation is less than 3% (0.191 calculated against 0.186 observed).
The value of one of the coefcients was determined only once though it could have been determined
twice. Instead, they determined the value of another coefcient three times.
13
The values of two coefcients vary in about 10% between their determinations. Yet since their values
are one order of magnitude smaller than the other two, an error of 1% in the values of one of the others
can lead to an error of 10% in the smaller coefcients. Thus, Riecke and Voigt could regard the
conrmation as satisfactory.
12
ARTICLE IN PRESS
592
regarded this determination as an important task of their piezoelectric experiment,

an aim independent of the conrmation of the theory. The task turned out to be
more complicated than they expected. As in the previous experiments, they read the
voltage in the laboratory. Voigts piezoelectric moduli, however, related elastic
pressure or stress to polarization. Polarization was known to be equal to the charge
surface density, which could be calculated from the voltage and the capacity of the
system. Like their predecessors, they used an external condenser, whose capacity was
calculated from its dimensions, but in a somewhat different way from both Czermak
and the Curies. Voigt and Riecke connected the condenser to the apparatus and
compared voltage readings due to the piezoelectric effect with and without the
condenser, nding the ratio between the capacity of the system in the two states.
From their ratio and the capacity of the condenser, they calculated the capacity of
the apparatus (1892, pp. 534, 544545). With its value, they found that
d11 5:31 108 statcoulomb/dyne for quartz and d33 4:70 108 for tourmaline (Riecke & Voigt, 1891, p. 254).
Their experimental method was closer to Czermaks than to the Curies. In
measuring coefcients of quartz and tourmaline, the Curies determined only the
electric charge due to an additional weight on a known condenser by a null deviation
of an electrometer needle. Their method bypassed the measurement of the capacity
of the system and translation of a deviation of an electrometer dial to voltage.
Czermak and Riecke and Voigt, on the contrary, measured and accounted for all the
magnitudes involved in their experiment. They measured the capacity of the system
rather than bypassing it, and they read a deviation in the electrometer dial and
compared it with a standard cell rather than perform a null experiment. Czermak
tried to account for all parts of the apparatus, but less rigorously. For example, in
measuring the capacity of the system, he neglected the small capacity of the
electrometer. Moreover, Riecke and Voigt reduced systematic and accidental sources
of errors mathematically. They accounted for the divergence from the correct
position of their electrometer dial (an error that did not arise in the Curies setting);
they carried out observations under several conditions; and they used the method of
least squares for error analysis. Riecke and Voigts experiment was more meticulous
than that of the Curies because they examined every detail of their apparatus.
Yet whether it was really more accurate was a matter of opinion. They surely
believed with most of the German physicists that their experiment would yield more
precise and accurate results. German measuring physicists thought that one should
account for external inuences on experimental results rather than bypass them.
Among the students of Neumann like Voigt, this was even clearer. Neumann and
his seminar students did not believe that redesigning an instrument could reduce all
errors. In their view, it would be futile to pursue the material perfection of
instruments over the accurate mathematical determination of errors (Olesko, 1991,
p. 302).
An 1860s controversy about the appropriate way to reach exact and valid results
between Pape, another member of Neumanns school, and the inuential French
experimentalist Regnault illustrates their attitude. Would accuracy be achieved by
noting all errors and accounting for them through mathematical analysis, as Pape
ARTICLE IN PRESS
593
claimed, or through elimination of errors by an experimental arrangement that

would permit their neglect, as Regnault claimed? Regnault considered mathematical
techniques unsuitable for examination of experiments because they were too far from
the reality of the laboratory. With his so-called methode directe Regnault tried to
obtain data directly varying experimental apparatuses and arrangements a priori
.
rather than calculating errors of experimental data a posteriori (Dorries,
1998a,
p. 124). While Pape favored carrying out experiments under various conditions,
Regnault preferred comparing results of experiments performed under the same
conditions. Errors, except accidental errors, were to be reduced in the experiment
itself by modifying the setting, the apparatus and the target according to the
judgement of the experimentalist, not in its data analysis. Pape pointed out that
Regnault was and would be unable to avoid signicant experimental errors. Pape
thought that by accounting mathematically for errors, a scientist could reach more
accurate and valid results than in an attempt to eliminate them. The scientist could
not rely on the personal mastery of the experimentalist but on physical-mathematical
.
.
analysis (Olesko, 1991, pp. 378382; Dorries,
1998b, p. 260). Dorries
also noted that
Regnaults whole measuring enterprise became artistic to an extent where subjective
judgement prevailed at the expense of objectively reproducible measurements
(1998a, p. 137).
Pape and Regnault present two attitudes towards experimentation: a mathematical
versus an artisanal approach. The former used common procedures in locating
possible errors and reducing them; the latter employed more particular solutions to
specic experimental settings and their modication. They resemble analytic and
synthetic geometry. Nothing in the artisanal approach was intrinsically more artistic
than synthetic geometry. Experimentation in both approaches involved theoretical
considerations and assumptions about the working of the apparatus. In the artisanal
approach these were primarily applied in designing the experiment before its
performance while in the mathematical, primarily after it was performed.
Like Pape, Voigt, and Riecke displayed a high degree of condence in their
analysis of experimental instruments and their systematic and accidental errors. Like
him, they preferred the mathematical analysis of all the experimental components to
bypassing measurements. Other German measuring physicists revealed the same
tendency. Data analysis and mathematical theory of measuring instruments and
their deviations were central to both Neumanns and Webers schools. This centrality
indicates that German measuring physicists trusted their method of analyzing
complicated results more than experimental methods for bypassing complexities.14
14
For example, Friedrich Kohlrausch, who was an inuential teacher of Riecke (Voigt, 1915) considered
calculation of the effect of the earths magnetic moment necessary for exact measurement of electric
resistance standards. The British, aware of the difculties involved [in the magnetic measurements], tried
to develop alternative methods that would bypass it. Kohlrausch did not (Olesko, 1996, p. 138). Oleskos
work demonstrates the centrality of data analysis for German physicists in both Webers and Neumanns
schools. Olesko reveals differences between German and British attitudes towards practices of exact
measurements. The examples of the Curies experiment and Regnault indicate that the French were closer
to the British than to the Germans in their preference for bypassing the complicated mathematical analysis
of extraneous inuences.
ARTICLE IN PRESS
594
Czermak, who belonged to neither of these schools but was educated as a theoretical
physicist in the German sphere, displayed a similar approach in his experiment,
although he put less emphasis on error analysis. The Curies, on the other hand,
proved themselves true and successful followers of Regnaults method. By
appropriate design of their apparatus, they measured the sought magnitude almost
directly, without a need for elaborate data analysis and mathematical elimination of
error. Jacques Curie continued using this method (reaching excellent results) well
after Riecke and Voigts experiment. This example suggests that French physicists
.
were right in acknowledging Regnaults lasting inuence (Dorries,
1998b, p. 260).
Elisabeth Garber claims that the disciplinary borders between experimental
physics and mathematics were more rigid and dened in France than in Germany
and Britain. Experimental physicists in France [she further claims] regarded
their work as purely empirical and devoid of all hypotheses (Garber, 1999, p. 314).
The German measurements in piezoelectricity did not involve more hypotheses
than the French, but clearly did apply more mathematics. If Garber is right
about the status of mathematics in French physics, its consequences should be
most conspicuous in quantitative measurements, as suggested by the example of
piezoelectricity.
5.3. Determining the constants

.
Riecke and Voigts results, presented to the Gottingen
Society for Science on
August 1, 1891, did not agree with those of the Curies. They were lower by 16% (for
quartz, 5.31 compared to 6.32) and 13% (for tourmaline, 4.7 compared to 5.4) than
the Curies.15 In the published version of the communication, dated November 11,
these results were given alongside those of the Curies, without any further comment
(1891, pp. 254255). Yet before they sent a more complete report of their experiment
to the Annalen der Physik, dated autumn 1891, they realized that the value
calculated for the capacity of the system given earlier was however surely too
small (1892, p. 545). Thus, Riecke and Voigts gures for the moduli were also too
small.
The estimation of the capacity was too small, they explained, because shellac on
the external condenser increased its capacity and was not accounted for by the
calculation. Moreover, the particular state of the condenser did not permit a
theoretical calculation of the capacity. They had to nd it experimentally. They
used an electromagnetic method of alternating current suggested by Maxwell
in his Treatise of Electricity and Magnetism (Maxwell, 1892, pp. 420425).
Here, the capacitor was connected by a tuning fork to a battery that charged it
and a galvanometer through which it was discharged. This method required a
special series of complicated measurements to nd the capacity of the external
15
Riecke and Voigt quoted 6.3 and 5.3, respectively. I presume that they did not attribute to the Curies
experiment higher precision than that (the Curies themselves published a value of 6.32 in 1889). The error
in the latter number is probably due to a simple mistake.
ARTICLE IN PRESS
595
condenser.16 With the new value for the condenser, Riecke and Voigt (1892, pp. 546
549) recalculated the capacity of the experimental apparatus. The long procedure
required to determine the capacity is a good example of the effort needed for exact
measurements.
Only after this lengthy procedure could they calculate the absolute magnitudes of
the piezoelectric moduli of quartz and tourmaline. Their values were now little
higher than those J. and P. Curie had found a decade earlier. For d11 of quartz they
found 6.45 versus 6.32; for d33 of tourmaline, they found 5.71 compared to 5.4,
all multiplied by 108 : This time, Riecke and Voigt stated that the smaller
discrepancy might well be especially due to that we took into our account loss of
electricity that took place during the observation. In Hrn. Curie not the least remark
relating to this is found (1892, pp. 549550). In particular, they probably had in
mind the corrections they made in the reading of the electrometer. Yet, their critique
was not limited to a specic procedure. It also revealed their uneasiness with the brief
and general descriptions of the Curies experiments, in which only the end results,
but not the values of the various observations, were given. Riecke and Voigt
probably shared Wilhelm Webers ideal of an experimental account. Weber argued
that the surety and certainty of the result was founded in good part on the
investigators thoroughness in reporting the experiments instruments, how they were
used, how the trials were performed and what changes were made, and how the data
was taken and reduced (Olesko, 1996, p. 119). The Curies reports lacked this
thoroughness and thus did not give German physicists the desired certainty.17
According to Riecke and Voigt, their earlier determination of the condensers
capacity was surely too small because the calculation disregarded the accidental
shellac that increased its capacity. Thus, they implied that no external stimulus was
required to convince them of the need to re-examine the capacity that was surely
wrong. However, this was not obvious to them when rst analyzing the results of the
experiment. Even in a public announcement of its results, and in its later written
version, they gave results based on surely too small capacity. Only a close reexamination of the apparatus, including the condenser, made it clear that the earlier
values were wrong. A new set of measurements to replace those made on a mistaken
16
The tuning fork alternates between the battery and the galvanometer, acting as oscillating switch. At a
certain frequency of oscillation (in their experiment, 31.47 times a second) the needle of the galvanometer
is steady. To measure this frequency exactly, they used a special apparatus with photographic paper. The
capacity is then expressed as a function of the resistance of the galvanometer, the frequency of steady
current, and the ratio between the current that discharges the condenser and that in a branch circuit in
which a resistance of 22,000 O was connected to the battery. (The measurement of the current in the
branch circuit actually indirectly gives the voltage of the cell, which is not expressed directly by the
formula. In this way they measured the voltage of the pile rather than relying on their book values as the
Curies did with the same kind of Daniell cell.) The last two were measured in the experiment. They
made about 20 measurements of the two currents with and without the measured condenser. In each case
they calculated the capacity of the system. From these measurements, they obtained the mean values of the
capacities of the tuning fork circuit, and of the circuit without the condenser. The difference between
those two values gave the capacity of the condenser.
17
The presentation of the Curies results also leaves much to be desired by the historian, especially
considering the absence of their laboratory notes.
ARTICLE IN PRESS
596
premise was necessary. Evidently a surface examination of the condenser was not
part of the procedure which they applied in this experiment.
But why did Riecke and Voigt take the trouble to return to the laboratory and start
inspecting apparatus that they had already checked, and from which they had even
published experimental results? I think that the answer lies in the disagreement with
the earlier results of the Curies and, to a lesser degree, those of Czermak.18 In spite of
their greater trust in their own method, Riecke and Voigt did not dismiss the Curies
method and resultsresults conrmed, though unsatisfactorily, by Czermak.19 The
higher accuracy that they attributed to their own method was not sufcient to account
for the large gap of 16% between the results. The errors accounted for by their analysis
were much smaller than this gap, which suggested a fault in one of the experiments.
Since Riecke and Voigt could not examine the Curies apparatus, they probably
decided to re-examine their own. An error was indeed found in the value ascribed to
the capacity. It is common to re-examine experiments and even to perform them again
in cases of discrepancy with a theory. This was not the case here, since the piezoelectric
theory was indifferent to the magnitude of the moduli. Here the results of an earlier
experiment offered grounds for comparison and thus inuenced the results of the later
experiment.20 It is unlikely that Riecke and Voigt would have noticed a little shellac on
a condenser had their results agreed with those of the Curies or had the Curies not
performed any previous measurements.
5.4. Examining the existence of direct pyroelectricity
Correcting the values of the moduli was important not only for the records and
agreement with the Curies results but also for inferring the existence of a genuine
direct pyroelectric effect. Following others, Voigt assumed in his general theory that
pyroelectricity is an electric effect of strain, due to thermal deformation of the
crystals, rather than a direct thermal effect. Thus the theory expressed the electric
effect of temperature change as a function of the coefcient of thermal expansion
and the piezoelectric constants (eij ). If pyroelectricity was only a secondary
phenomenon of piezoelectricity, the constant calculated from these two coefcients
would be equal to the one obtained from direct measurements. Riecke and Voigt
compared the calculated constant from this experiment with one obtained by Riecke
from direct measurements on another specimen a year earlier. Adjusting Rieckes
results to the conditions of the piezoelectric experiment, they found that the
calculated indirect effect was 9% larger than the observed effect (1.34 statcoulomb/
cm2 degree compared to 1.23). Considering that a number of constants were involved
18
The published sources are silent about the reason for the re-determination. I failed to locate any
manuscript that illuminates the question.
19
Riecke and Voigt did not mention Czermaks determination, but they knew the paper (Voigt quoted
results from it as early as 1890; Voigt, 1890). They probably disregarded Czermaks result because they
depended on his mistaken theory. Yet, despite the problematic determination of Czermak, one could use
the measurements that he did not need to modify (those on the plates with parallel surfaces) to reach
results that are close to the Curies.
20
It should be clear that the contingent historical way in which the experimental results were attained
does not alter the validity of the methods eventually used, or their results.
ARTICLE IN PRESS
597
in the determination of the theoretical value, they concluded that the agreement
between the results conrmed the basic assumptions of the theory (1892, pp. 550
552). Interestingly, the correction of the piezoelectric moduli did not change their
conclusion that pyroelectricity is a secondary effect of piezoelectricity. They also
found reasonable agreement with their earlier mistaken values, though these were
lower by 20% than the corrected values. Then they cited a value somewhat lower
(1.18) for the observed pyroelectric constant and reached a deviation of less than
10% (their calculated value was 1.08; Riecke & Voigt, 1891). The arrival at the same
expected conclusion from two different sets of values revealed exibility in analyzing
precise qualitative experimental results.
In 1898, Voigt returned to empirically examining the existence of a direct
pyroelectric effect in tourmaline. It was still possible that an effect smaller than the
experimental error existed. It was more likely because the theoretical and the
observed value deviated by 9%. Furthermore, Riecke and Voigt had not measured
all the coefcients involved in the theoretical calculation of indirect pyroelectricity
on the same specimen. In the new determination, Voigt measured all the relevant
coefcients on the same specimen. This enabled him to compare the calculated
indirect effect with the observed one before determining their absolute values. Since
he found a signicantly lower value for the thermal expansion and therefore for the
indirect effect, the new experiment supported the existence of genuine direct
pyroelectricity (Voigt, 1898; Katzir, 2001, pp. 214216).
6. Pockels determination of piezoelectric constants
Voigts study of the relationship between the kindred phenomena of piezo- and
pyroelectricity showed the necessity of exact measurements. Four years earlier, his
former student, Friedrich Pockels, had carried out a series of precise measurements
of constants to study the relation between piezoelectricity and electro-optics. Electrooptics is the effect of an electric eld on double refraction in crystals that changes the
polarization of light transmitted through the crystals. Pockels carried out
measurements to determine whether a direct electro-optic effect existed. In 1883,
.
Rontgen
and Kundt separately explained this effect as resulting from converse
piezoelectricity; i.e., a stress induced by an electric eld that produced an optic effect
.
through the already known piezo-optic effect (Rontgen,
1883; Kundt, 1883). In 1890,
Pockels formulated a mathematical theory that accounted for electro-optics. He
showed that the joint effects of piezoelectricity and piezo-optics accounted
qualitatively for electro-optics, but he left open the possibility of a direct electrooptic effect (Pockels, 1891). To decide whether such a direct effect existed, it would
be necessary to examine whether the joint effects accounted quantitatively for the
whole observed effect of electro-optics. The coefcients of the three phenomena and
of elasticity had to be known for this experiment. Therefore, Pockels had to
determine their values. Like Voigt, he measured all the coefcients (except those of
elasticity, which he took from the literature) on the same specimens to make sure
that the results would not be inuenced by differences between specimens of the
same crystal (Pockels, 1894; Katzir, 2001, pp. 206214).
ARTICLE IN PRESS
598
Pockels studied three crystal species: quartz, sodium chlorate and Rochelle Salt
(also called Seignette salt). Since the piezoelectric moduli of the latter two were not
measured, and he used an unexamined quartz specimen, he measured the
piezoelectric moduli of all three. His measurement method was similar to Riecke
and Voigts. He measured deviations of the electrometer needle and compared them
with the readings of a Clark cell, but not before he had adjusted the reading to the
correct value using an expression he developed for the oscillations of the needle. The
capacity of the system was again determined by adding an external air condenser.
This time, the theoretical calculation of its capacity seemed valid (Pockels, 1894, pp.
6973, 131132, 137143).
Like Riecke and Voigt, for the measurements of quartz, he applied pressure in one
plane but in only three directions. With such a small number of observations, he
could not analyze the data statistically but calculated the values from the expressions
of the polarizations in each case. With three equations for two variables and without
an explicit method to nd averages, his determination of the values of the moduli
were somewhat arbitrary. Riecke and Voigts use of the method of least squares to
determine the value of the moduli made their determination superior to any arbitrary
direct calculation. Pockels justied the values he obtained by comparing the
observational data with gures calculated with these values. He found an agreement
within 4% (1894, pp. 131144). Pockelss results agree with Voigts general theory,
but after Riecke and Voigts more thorough conrmation in a similar situation, this
corroboration added very little. Pockels, however, was not interested in the
conrmation of the theory, which he considered conrmed, but in the value of the
quartz moduli, which he needed in order to examine the relation between
piezoelectricity and electro-optics. For d11 ; Pockelss value is in excellent agreement
with that of the Curies, a difference of less than 1%; it is 3% smaller than Riecke
and Voigts (see Table 2). However, his value for d14 is considerably larger than
that found by Riecke and Voigt: 1.925 in comparison to 1.45, a difference of about
30%. Pockels did not explain this gap; he probably did not know how. Later
measurements agreed more with Pockelss, nding values even higher than his.
While Pockelss measurements of sodium chlorate largely conrmed the general
theory, the test of the theoretical relations for Rochelle salt was more problematic.
According to the theory, Rochelle salt has three independent moduli. Measuring the
electric effect related to one of these moduli (d14 ), Pockels found that the
electrometer dial did not stabilize but continued increasing in a way that precluded
him from nding its proper deviation. Instead of a xed value, he could only state a
range between 340 and 1180 108 (statcoul/dyne) for d14 ; concluding that its value
was left undetermined (1894, pp. 183189).21 The values that Pockels got for the
21
In hindsight, one can conclude that Pockels probably observed for the rst time an effect of ferroelectricity. Indeed d14 is not constant in Rochelle salt. Though Pockelss experiment left an open question
about the piezoelectric behavior of the crystal, he did not suggest it as a subject for further research. Its
experimental examination was only resumed in the 1920s by Valasek, who had formerly worked with
Voigt. A decade later, physicists in the USSR realized that Rochelle salt exhibits an unrecognized electric
phenomenon that they named after the crystal, Seignette-electricity, which is today usually called ferroelectricity since it is analogous to ferromagnetism.
Czermaks
Riecke and Voigts
Pockels
Crystals and coefcients

measured
Quartz (d11 ), tourmaline

(d33 )
Quartz (d11 )
Quartz (d11 ; d14 ),

tourmaline (d15 ; d22 ; d31 ; d33 )
Value of quartzs d11

(in statcoul/dyne 108)
Number of different settings
by which it was determined
Measurement of systems
capacity
Determination of a
capacitors capacity
Voltage reading
6.32
6.27
First 5.31
Second 6.45
5
Quartz (d11 ; d14 ), Rochelle

Salt (d14 ; d25 ; d36 ), sodiumchlorate (d14 )
6.27
No
Yes
Yes
Yes
Theoretical
Theoretical
Theoretical
Null experimentneedle at
zero
No
Deviation compared with a

Daniell cell
No
First theoretical, Second

experimental
Clark cell
For quartz
Statistical analysis of the

results

Clark cell
No
ARTICLE IN PRESS
Curies
Table 2
Comparing the four experimental determinations of piezoelectric coefcients
599
ARTICLE IN PRESS
600
modulus d14 contradict the assumption that it is a constant. Still, he continued to

refer to the coefcient as an undetermined constant, implying that someone would
eventually be able to determine its constant value. He suggested an ad hoc
explanation for his inability to observe a constant value and thus for the apparent
contradiction with piezoelectric theory.22 Rochelle salts other two moduli did show
a constant behavior. Apparently Pockelss experiment contradicted the assumption
of the general theory that the piezoelectric effect is steady and linear, at least in
Rochelle salt. Nevertheless, this contradiction did not raise any questions about the
validity of Voigts theory.
The conrmation of Voigts theory (and its many predictions for various classes of
crystal and various applications of stress) was very limited.23 The two experimental
studies by Riecke and Voigt and Pockels were the only tests of the theory undertaken
in the 1890s. They conrmed the theory for only three species, which represented
only three classes of crystals out of the 20 that should show piezoelectric behavior
according to Voigts theory. Of the various cases studied in Voigts general theory,
they examined only unidirectional pressure. Evidently, they saw no need for a
conrmation of Voigts predictions for various geometrical circumstances, since they
were derived mathematically from the basic equations of the species, whose
conrmations were the subject of their research. This limited examination reveals the
by then virtually universal condence in mathematical deductions in physics. Despite
the limited conrmation, the validity of the theory was not put in doubt, and no one
called for additional work. Apparently the scientic community regarded it as
satisfactory.24
The limited effort to conrm the theory was surely related to its success in
accounting for all earlier (mostly qualitative) empirical data and to condence in its
basic assumptions. The considerations of symmetry on which the theory was based
were grounded in many phenomena, and the linearity of piezoelectricity had been
repeatedly observed. Still, the linearity was checked in only a few species, as was the
validity of symmetry considerations for these phenomena, and the measurements on
Rochelle salt seemed to contradict even the constancy of the effect for the same
stress.
22
Pockels suggested that the source of the continuous rise in electric tension measured by the
electrometer might be in the magnitude of the effect, which is larger by two orders of magnitudes than
those previously observed. Therefore, he conjectured that the connection to the ground and the
electrometer do not supply enough electric charge to the metal plates attached to the crystal. In such a
case, their electric eld is smaller than the inner electric eld in the crystal, so a net electric eld remains in
the crystal. This eld produces electric current (in the crystal) that raises the tension between the metallic
plates. He also suggested an experimental arrangement to examine the explanation, but as far as I know,
neither he nor anyone else tried to construct it.
23
Indeed many of these predictions were un-testable, but the examination of many others (like the
behavior under unidirectional pressure of various species) was feasible.
24
I have not found any comment on the need to conrm the theory after 1892 (except concerning the
side question of the genuine existence of pyroelectricity). The participants themselves referred in other
places to the empirical examination of the theory as extensive (Voigt, 1894, p. 343; Pockels, 1895, 1905,
p. 780).
ARTICLE IN PRESS
601
Pockels measurements of the piezoelectric, piezo-optic and electro-optic

coefcients in Rochelle salt did not permit a clear conclusion about the existence
of a direct electro-optic effect in that crystal. Yet the results in quartz and sodium
chlorate gave clear evidence for the existence of such an effect. Thus, Pockels (1894,
p. 204) concluded, electrostatic forces [elds] exert [ausuben]
a direct inuence on
.
the transmission of light in piezoelectric crystals. Research on the relations between
the kindred phenomena of piezoelectricity, pyroelectricity and electro-optics revealed
the fertility of measuring physics. Only precise quantitative measurements whose
design and interpretation were based on the mathematical theory of the effects could
tell whether the last two were genuine phenomena. In 1909, in a lecture about the
struggle for the decimal in physics, Voigt asserted that this kind of research raised
questions that could be neither known nor resolved without precise measurements
(Voigt, 1909).25 He did not mention the rather complicated examples from the study
of piezoelectricity, but his and Pockels experiments were excellent examples for his
.
claim. Though Rontgen
and Kundt had observed electro-optic effects as early as
1883, they did not realize that they had observed a new, genuine phenomenon. They
did not even raise the question of whether such a phenomenon existed, as the
theoretically oriented Pockels did seven years later. He later succeeded in answering
this question with precise measurements. In the study of pyroelectricity, Voigt
showed that a more thorough and accurate experiment in which all the variables
were examined could reveal a phenomenon that earlier experiments overlooked.
7. Conclusions
The experimental conrmation of piezoelectric theory and the determination of its
constants represented the main functions of measuring physics at the end of the
nineteenth century. Clearly testing mathematical theories was one function, though
apparently not so central as one might have expected. Physicists were satised with
few experimental conrmations limited to a few cases. Collecting information about
nature by nding the precise mathematical coefcients that characterized physical
processes was another function. The precise determination of a constant to the
highest degree possible was not only an end in itself but also a means to reveal
relations between phenomena and to discern whether genuine physical interactions
existed. Pockels and Voigt were concerned with the particular (e.g., a constant of a
specic material) mostly as a means to reach the general (e.g., the existence of a
direct electro-optic effect). Precise measurements, as Voigt claimed, raised and
answered questions inaccessible through qualitative experiments (Voigt, 1909, p. 73).
Exactitude was conceived as an indispensable means to a superior end.
25
Such precise measurements were not a monopoly of measuring physicists. As the example of the
Curies shows clearly, this was not so in France. Yet, in Germany, experimental physicists also occasionally
carried out exact measurements. The question of the existence of a direct pyroelectric effect was examined
.
.
in the 1910s by W. Rontgen,
one of the leading experimentalists of the time (Rontgen,
1914). Yet during
.
the 1880s, Rontgen
performed only qualitative experiments on piezoelectricity, and he did not study the
eld at all in the 1890s.
ARTICLE IN PRESS
602
Considerable effort to attain high precision and eliminate even small experimental
errors was evident in all four determinations of piezoelectric constants discussed
here. Apparently, the six scientists trusted the precision of their determinations. They
all shared an ethos of exactitude. Their measurements were based on
mathematical laws, which gave the constants meaning. Still, they presented different
methods of reaching exactitude. While the Curies bypassed potentially complicated
determinations and eliminated any signicant source of error, the German
measuring physicists preferred to measure all quantities and base the elimination
of error on theoretical-mathematical analysis.
In the tradition of Regnault, the Curies reduced potential errors in their
measurements by manipulating the apparatus. German measuring physicists
regarded the mathematical method of reducing errors as superior to other methods.
They did not adopt the Curies earlier method of null experiment, nor did the Curies
adopt theirs. Today, the mathematical analysis of error has become a virtually
inseparable part of experimentation. Yet, the choice between experimental designs
that bypass some measurements and those that do not still exists. Years later,
Jacques Curies null experiment was regarded as most reliable, more than those of
Riecke, Voigt, and Pockels (Cady, 1964, pp. 217218).
The two attitudes towards precise measurements characterized two different
traditionsthat of French experimentalists and that of German theoreticians.
However, nationality was not the decisive factor in the difference between the two
attitudes. Afliation with a particular school or tradition was more important.
Indeed, these schools were each related to a specic national, regional, institutional,
educational, and linguistic setting. Weber and Neumann found their followers in
Germany; Regnault in France. Moreover, the tradition of measuring physics, with
its emphasis on mathematical reduction of errors, was probably unique to Germany.
Yet, these schools did not include all German or French physicists. Germany had its
own two separate traditions of experimental and measuring physics. Experimental physicists in Germany did not share the attitude of their compatriot
.
measuring physicists. In 1912, Conrad Wilhelm Rontgen,
a prominent representative of the experimental tradition, employed the method of null experiment to
.
determine the value of the quartz piezoelectric coefcient (Rontgen
& Joffe! , 1913).26
.
Like his French colleagues, Rontgen
attempted to reduce errors in the experiment
through a design of the apparatus (instruments, targets, and detector) rather than
through the mathematical error analysis employed by his compatriot theoreticians.
.
Rontgen
showed an inclination toward the artisanal rather than the mathematical
approach to precise measurements.
Riecke and Voigts research illustrates the complexity of the experiment and its
analysis. It also exhibits its exibility and the ability of the experimentalists to
modify and adjust it. In this case, the modication of the experiment did not involve
reconsideration of its theory but a fresh inspection of its equipment that revealed an
accidental technical problem. The experiment was modied by the addition of a
.
Rontgens
measurements deserve a separate discussion elsewhere. They are beyond the chronological
scope of this paper.
26
ARTICLE IN PRESS
603
complicated measurement of capacity. Consequently, the results were changed, but

not the other components of the experiment. This episode supports the claim of the
New Experimentalism that theory does not give complete directions for practical
experimental decisions.27 Theory could not tell Riecke and Voigt which part of the
experiment needed modication and how to modify it. The solution of the problem
was based more on experimental methods than on theoretical ones. For measuring
the capacity, they adopted the tuning fork method from Maxwells treatise.
Although regarded as the epitome of his electromagnetic eld theory, Maxwells
book was far from being a systematic presentation of the theory and was not limited
to its presentation. It was also a detailed presentation of experimental method.
Despite the methods reliance on more advanced theory than that needed for the
theoretical calculation of capacity, it did not require eld theory and its specic
concepts. Maxwells analysis of the charging and discharging of condensers was also
accepted by physicists who rejected his system. Riecke and Voigt themselves were not
Maxwellians.
Peter Galison has called historians attention to the freedom experimentalists have
in deciding when to end an experiment (Galison, 1987). One can always suggest
further measurements of components of the apparatus, the measuring instruments
themselves, or the target in order to check possible systematic and accidental errors.
Galison examined episodes from the history of particle physics in the twentieth
century; this story shows that ending an experiment was also sometimes a long
and complicated process in the study of bulk matter at the end of the previous
century. Experiments were not over once the measurements had been made. Neither
data analysis nor publication of results terminated the research. The Curies were
quick to pronounce their determination of the piezoelectric constants. Yet they
continued to measure them, returning to the same experiment for many years:
Jacques pronounced modied results for basically the same measurement almost 30
years after the initial publication. Voigts experiments on the existence of direct
pyroelectricity, with Riecke and alone, showed that a conclusion from an experiment
(not only a determination of a magnitude) can be changed a few years after its
performance. In 1898, theoretical concerns combined with dissatisfaction with the
precision and method of the former experiment (which did not measure all variables
on the same specimen), led Voigt to complete the experimental investigation that had
apparently been concluded in 1891.28 Like the Curies, Riecke and Voigt announced
their determination of piezoelectric moduli soon after they obtained numerical
results, but publication did not end the experiment. They returned to the laboratory
soon after to correct their ndings. This example shows that, in the absence of
theoretical expectations, earlier experimental results can inuence a new experiment;
they can even determine when it would end. Perhaps this inuence characterizes
27
For a statement of this claim, see Gooding (2000).

Voigts 1898 experiment might be seen as independent of the previous one, but since it was done on
the same specimen and since Voigt used the same piezoelectric measurements and re-analyzed the old
experiment with new data obtained in 1898, I maintain that both experiments can also be regarded as one.
This semantic choice can help to recognize that the analysis of the experiment and its conclusions can be
changed even years after they have found their way to a scientic journal.
28
ARTICLE IN PRESS
604
more measurements of constants than other experiments (e.g., those that test
theories). Further research is needed in order to corroborate this hypothesis.
Ending an experiment following a previous experiments results contains a
paradoxical element: the new experiment was designed, among other aims, to
examine exactly the correctness of the previous results, and to nd more accurate
and valid results. Nevertheless, the circularity is only apparent. Although the ending
of the new experiment was done in view of the previous one, the justication for the
correction of the results and the procedures of their re-examination were
independent of the preceding results. True, the new experimental results were
inuenced by the previous ones, but were not determined by them. Results that
disagreed with previous data were published by Riecke and Voigt as long as no error
was found in the experiment and its analysis. Plausibly, without the Curies previous
results, the faulty results would have remained standing.29 However, the methods
and procedures used by Riecke and Voigt in measuring the correct capacity, and
their condence in them, provide excellent reasons to believe that had their results
led to disagreement with the Curies, they would not have hesitated to publish them.
Acknowledgements
This paper emerged from my Ph.D. dissertation on the early history of
piezoelectricity written at Tel Aviv University under the supervision of Ido Yavetz.
It is a pleasure to thank Ido for the many discussions we had, his critical comments
and his suggestions throughout my research. Part of the research for this paper was
carried while I was a visiting scholar at the Department of History and Philosophy of
Science at University of Pittsburgh. I wish to express my gratitude for its hospitality.
It is a special pleasure to thank John Norton for our discussions of my work
and his suggestions, many of which found their place in this paper. The library of the
Max Planck Institute for the History of Science in Berlin kindly sent me
indispensable sources for this research, and I am pleased to express my gratitude
to its staff, especially to Matthias Schwerdt. This paper beneted from the comments
of Leo Cory and the linguistic editing of Anne Sartiel; thanking both is a pleasure.
I would like also to thank the anonymous referees for their valuable criticism and
advice.
References
Cady, W. G. (1964). Piezoelectricity: An introduction to the theory and applications of electromechanical
phenomena in crystals. New York: Dover Publications, Inc.
Curie, J., & Curie P. (1880). D!evelopement, par pression, de l!electricit!e polaire dans les cristaux h!emi"edres
a" faces inclin!ees. In P. Curie, Oeuvres de Pierre Curie, 1908 (pp. 69). Paris: Gauthier-Villars.
29
Since Riecke and Voigt needed to measure the capacity of the specic condenser used in the
experiment, they probably would not have found this fault a few years later, when further measurements
were made.
ARTICLE IN PRESS
605
Curie, J., & Curie P. (1881a). Lois du d!egagement de l!electricit!e par pression dans la tourmaline. In P.
Curie, Oeuvres de Pierre Curie, 1908 (pp. 1517). Paris: Gauthier-Villars.
Curie, J., & Curie P. (1881b). Les cristaux h!emi"edres a" faces inclin!ees comme sources constances
d!electricit!e. In P. Curie, Oeuvres de Pierre Curie, 1908 (pp. 2225). Paris: Gauthier-Villars.
Curie, J., & Curie, P. (1882). Ph!enom!enes e! lectriques des cristaux h!emi"edres a" faces inclin!ees. Journal de
Physique th!eorique et appliqu!ee, 1, 245251.
Curie, J., & Curie, P. (1889). Dilatation e! lectrique du quartz. In P. Curie, Oeuvres de Pierre Curie, 1908
(pp. 3555). Paris: Gauthier-Villars.
Curie, M. (1910). Trait!e de radioactivit!e, Vol. 1. Paris: Gauthier-Villars.
.
Czermak, P. (1887). Uber
der elektrische Verhalten des Quarzes. Sitzungsberichte der MathematischNaturwissenschaftlichen Classe der Kaiserlichen Akademie der Wissenschaften, Wien, 96, 12171244.
.
Dorries,
M. (1998a). Vicious circles, or, the pitfalls of experimental virtuosity. In M. Heidelberger, & F.
Steinle (Eds.), Experimental essaysVersuche zum Experiment (pp. 123140). Baden-Baden: Nomos.
.
Dorries,
M. (1998b). Easy transit: Crossing boundaries between physics and chemistry in mid-nineteenthcentury France. In C. Smith, & J. Agar (Eds.), Making space for science (pp. 248262). Houndmills:
Macmillan.
Galison, P. (1987). How experiments end. Chicago: University of Chicago Press.
Garber, E. (1999). The language of physics: The calculus and the development of theoretical physics in
Europe, 17501914. Boston: Birkhauser.
Gooding, D. C. (2000). Experiment. In W. H. Newton-Smith (Ed.), A companion to the philosophy of
science (pp. 117126). Oxford: Blackwell.
Jungnickel, Ch., & McCormmach, R. (1986). Intellectual mastery of nature: Theoretical physics from Ohm
to Einstein. Chicago: University of Chicago Press.
Katzir, S. (2001). A history of piezoelectricity: The first two decades. Ph.D. dissertation, Tel Aviv
University.
Kuhn, T. S. (1977). Mathematical versus experimental traditions in the development of physical science. In
T. S. Kuhn (Ed.), The essential tension: Selected studies in scientific tradition and change (pp. 3165).
Chicago: University of Chicago Press.
Kundt, A. (1883). Ueber das optische Verhalten des Quarzes im electrischen Felde. Annalen der Physik und
Chemie, 18, 228233.
Maxwell, J. C. (1892). A treatise on electricity & magnetism (3rd ed.). Oxford: Clarendon Press.
Olesko, K. M. (1991). Physics as a calling: Discipline and practice in the Konisberg
seminar for physics.
.
Ithaca: Cornell University Press.
Olesko, K. M. (1996). Precision, tolerance, and consensus: Local cultures in German and British resistance
standards. In J. Buchwald (Ed.), Scientific credibility and technical standards in 19th and early 20th
century Germany and Britain. Dordrecht: Kluwer Academic Publishers.
Pockels, F. (1891). Ueber die Aenderungen des optischen Verhaltens und die elastischen Deformationen
dielektrischer Krystalle im elektrischen Felde. Neues Jahrbuch fur
. Mineralogie, Geologie und
PalaeontologieBeilage-band, 7, 201231.
Pockels, F. (1894). Ueber den Einuss des elektrostatischen Feldes auf das optische Verhalten
pi.ezoelektrischer Krystalle. Abhandlungen der koniglichen
Gesellschaft der Wissenschaften zu Gottingen,
.
.
39, 1204.
Pockels, F. (1895). Pyro- und Pi.ezoelektrizit.at. In A. Winkelmann (Ed.), Handbuch der Physik (1st ed.),
Vol. 3(part 2) (pp. 527550). Breslau: Eduard Trewendt.
Pockels, F. (1905). Pyro- und Pi.ezoelektrizit.at. In A. Winkelmann (Ed.), Handbuch der Physik (2nd ed.),
Vol. 4(part 1) (pp. 766793). Leipzig: Johann Ambrosius Barth.
.
Poggendorffs, J. C. (1945). Biographisch-literarisches Handworterbuch,
Vol. 4. Ann Arbor, MI: J.W.
Edwards.
Riecke, E. (1886). Ueber die Pyroelectricit.at des Turmalines. Annalen der Physik und Chemie, 28, 4380.
Riecke, E., & Voigt, W. (1891). Die Pi.ezoelectrischen Constanten des Quarzes und Turmalins. Gottingen
.
Nachrichten, 247255.
Riecke, E., & Voigt, W. (1892). Die Pi.ezoelectrischen Constanten des Quarzes und Turmalines. Annalen
der Physik und Chemie, 45, 523552.
ARTICLE IN PRESS
606
.
Rontgen,
W.C. (1883). Ueber die durch electrische Kr.afte erzeugte Aenderung der Doppelbrechung des
Quarzes. Annalen der Physik und Chemie, 18, 213228, 534551.
.
Rontgen,
W. C. (1914). Pyro- und pi.ezoelektrische Untersuchungen. Annalen der Physik, 45, 737800.
.
.
Rontgen,
W. C., & Joff!e, A. (1913). Uber
die Elektrizit.atsleitung in einigen Kristallen und uber
.
den
EinuX der Bestrahlung darauf. Annalen der Physik, 41, 449498.
Voigt, W. (1890). Allgemeine Theorie der pi.ezo- und pyroelectrischen Erscheinungen an Krystallen.
Abhandlungen der koniglichen
Gesellschaft der Wissenschaften zu Gottingen,
36, 199.
.
.
Voigt, W. (1894). Pi.ezo- und Pyroelectricit.at, die Electrische Inuenz und Electrostriction bei Krystallen
ohne symmetriecentrum. Gottingen
Nachrichten, 343372.
.
Voigt, W. (1898). L.asst sich die Pyroelectricit.at der Krystalle vollst.andig auf pi.ezoelectrische Wirkungen
zuruckf
. uhren?
.
Annalen der Physik und Chemie, 66, 10301060.
Voigt, W. (1909). Die Kampf die Dezimale in der Physik. Deutsche Revue, 34, 7185.
Voigt, W. (1915). Eduard Riecke als Physiker. Physikalische Zeitschrift, 16, 219221.

Measuring Constants of Nature Confirmation and Determination in Piezoelectricity

Încărcat de

Informații document

Drepturi de autor

Formate disponibile

Partajați acest document

Partajați sau inserați document

Opțiuni de partajare

Vi se pare util acest document?

Este necorespunzător acest conținut?

Drepturi de autor:

Formate disponibile

Measuring Constants of Nature Confirmation and Determination in Piezoelectricity

Încărcat de

Drepturi de autor:

Formate disponibile

PERGAMON

Studies in History and Philosophy of

Measuring constants of nature: confirmation

S. Katzir / Studies in History and Philosophy of Modern Physics 34 (2003) 579606

quantitative measurement using precision instruments had become a distinctive and

Discovery of the phenomenon by the Curies

Voigts general theory of piezoelectricity

Riecke and Voigts conrmation and

results depended on the material apparatus, various levels of theoretical and

2. First determination of the coefcients

S. Katzir / Studies in History and Philosophy of Modern Physics 34 (2003) 579606

Fig. 1. From Curie and Curie (1882).

S. Katzir / Studies in History and Philosophy of Modern Physics 34 (2003) 579606

Fig. 2. From the German translation of Curie (1910).

3. Czermaks determination of the quartz coefcient

its relationship to the phenomenon of pyroelectricitythe generation of electric

Fig. 3. From Czermak (1887).

S. Katzir / Studies in History and Philosophy of Modern Physics 34 (2003) 579606

Because he compared each reading of the voltage to that of a Daniell cell,

S. Katzir / Studies in History and Philosophy of Modern Physics 34 (2003) 579606

5. Voigt and Rieckes determination of piezoelectric moduli

S. Katzir / Studies in History and Philosophy of Modern Physics 34 (2003) 579606

S. Katzir / Studies in History and Philosophy of Modern Physics 34 (2003) 579606

regarded this determination as an important task of their piezoelectric experiment,

claimed, or through elimination of errors by an experimental arrangement that

S. Katzir / Studies in History and Philosophy of Modern Physics 34 (2003) 579606

5.3. Determining the constants

S. Katzir / Studies in History and Philosophy of Modern Physics 34 (2003) 579606

S. Katzir / Studies in History and Philosophy of Modern Physics 34 (2003) 579606

Riecke and Voigts

Crystals and coefcients

Quartz (d11 ), tourmaline

Quartz (d11 ; d14 ),

Value of quartzs d11

Quartz (d11 ; d14 ), Rochelle

Deviation compared with a

First theoretical, Second

Statistical analysis of the

Deviation compared with a

S. Katzir / Studies in History and Philosophy of Modern Physics 34 (2003) 579606

S. Katzir / Studies in History and Philosophy of Modern Physics 34 (2003) 579606

modulus d14 contradict the assumption that it is a constant. Still, he continued to

Pockels measurements of the piezoelectric, piezo-optic and electro-optic

S. Katzir / Studies in History and Philosophy of Modern Physics 34 (2003) 579606

complicated measurement of capacity. Consequently, the results were changed, but

For a statement of this claim, see Gooding (2000).

S. Katzir / Studies in History and Philosophy of Modern Physics 34 (2003) 579606

S. Katzir / Studies in History and Philosophy of Modern Physics 34 (2003) 579606

S-ar putea să vă placă și