Iournal of Microscopy, Vol. 180. Pt 2. November 199 5, pp. 140-147.

Received 5 November 1994; accepted 2 lune 1995

New polarized light microscope with precision universal

compensator

R . O L D E N B O U R G & G. ME1
Marine Biological Laboratory, Woods Hole, M A 02543, U.S.A.

Key words. Polarimetry, birefringence; modulator, liquid crystal.

Summary light microscope, we have developed a new type of polarized

light microscope ('new pol-scope'), which incorporates a
A new type of polarized light microscope ('new pol-scope')
precision universal compensator made of two liquid crystal
for fast and orientation-independent measurement of
variable retarders. The two variable retarders are computer
birefringent h e structure has been developed. The design
controlled and replace the traditional compensator of the
of the new pol-scope incorporates a precision universal
polarizing microscope. The new microscope optical set-up,
compensator made from two liquid crystal variable
plus a video camera and a specially designed, computerized
retarders. A video camera and digital image processing
image analysis system provide fast measurements of speci-
system provide fast measurements of specimen anisotropy
men anisotropy (retardance magnitude and azimuth) at all
(retardance magnitude and azimuth) at all points of the
points of the image constituting the field of view,
image forming the field of view. The images document fine
irrespective of birefringence axes. Because of its fast speed.
structural and molecular organization within a thin optical
high sensitivity and ease of use for measurements of optical
section of the specimen. The sensitivity of the current
anisotropies, the new instrument significantly advances the
instrument is 0.1 nm of specimen retardance, measured
analytical power of the polarizing microscope in all its
with data gathered in 0.43 s at all 640 x 480 image points.
traditional application areas.
Examples of birefringence measurements in biological
We note that the proposed measurement scheme is not
(microtubule arrays) and industrial (magneto-optical disc
limited to polarizing microscopy and it can be used in other
substrate) specimens are presented.
imaging or non-imaging applications involving precision
polarimetry. Therefore, we expect applications of the
proposed measurement scheme to evolve in other areas.
1. Introduction
including ellipsometry, remote sensing, material science.
Based on its great analytical power, the polarized light and manufacturing and processing control of plastic and
microscope has found numerous applications in fields glass materials.
such as biology, mineralogy, metallography, chemistry This article reports the first complete description of design
and for forensic studies (Chamot & Mason, 1958; Hart- and implementation of the instrument to measure linear
shorne & Stuart, 1960; Inoue, 1986; McCrone, 1991). retardances (birefringence) in a transparent microscopic
In biology, the polarized light microscope has the unique specimen (parts of the design concept and implementation
potential to measure submicroscopic molecular organ- were previously reported by us in a proceedings article; Mei
ization dynamically and non-destructively in samples & Oldenbourg, 1994; and in US patent application Serial
that, in general, can be kept in native environmental No. 08/241. 842). We first present the design outline
conditions. With the traditional polarized light microscope, including the polarization optical concept of the universal
however, single images display only those anisotropic compensator. Next, we discuss a specific algorithm to
structures that have a limited range of orientations with measure small specimen retardances at high spatial and
respect to the polarization axes of the microscope. temporal resolution concurrently for the whole field of
Furthermore, rapid measurements are restricted to a single view. This algorithm is then extended to include a
image point or single area that exhibits uniform birefrin- background correction procedure which corrects for
gence or other form of optical anisotropy (Allen et al., polarization aberrations introduced by different components
1963), while measurements comparing several image in the optical train, such as high-numerical-aperture (NA)
points take an inordinately long time (Inouk & Sato, 1966). lenses. With the background correction procedure the
To overcome the limitations of the traditional polarized measurement set-up is able to measure reliably specimen

ccj 199 5 The Royal Mlcroscop~calSoclety

 NEW P O L A R I Z E D L I G H T M I C R O S C O P E 141

Fig. 1. Schematic of the new polarized light micro-

scope. The optical design (left part) builds on the video camera
traditional polarized light microscope with the
conventional compensator replaced by two liquid

M
crystal variable retarders A and B forming the pre-
leftdrcuhr analyzer digital VO card
cision universal compensator. Retarder A is set at
a quarter wave retardance and is orientated with
its slow axis at 45" with respect to the linear objecke
polarizer. The combination of polarizer and retar-
der A produces left circularly polarized light. sample
Retarder B, set at a half wave retardance, flips
the polarization to right circularly polarized. The condenser
right circularly polarized light is effectively
blocked by the left circular analyser. Retardances variaMe retarder B, L12. 0"
A and B are varied by varying the voltages applied variable retarder A, 5/4.45°
to the liquid crystal devices using the retarder dri-
vers. Any given retardance from the specimen or
other elements in the optical train are compen- linear poiarizer OD
sated by adjusting retardances A and B without
interferencefilter
mechanically rotating any of the optical compo-
nents (see text). The video camera and electronic arc lamp
parts shown on the right of the figure provide
for image recording and analysis and for instru-
ment control.

retardance values below 0.1nm in magnitude. In the last of microtubule polymers that are formed at the poles of
section we describe the instrumentation and procedures the mitotic spindle in dividing animal cells). The images in
currently used to implement the new pol-scope. Fig. 2 were recorded with the set-up illustrated in Fig. 1 and
described in more detail in the Instrumentation section. In
the top left image of Fig. 2 the specimen is illuminated with
2. Design outline
right circularly polarized light. Astral rays appear bright,
The design of the new pol-scope builds on the traditional regardless of their orientation, against the dark background.
polarized light microscope introducing two essential mod- Background light is effectively blocked by the left circular
ifications: the specimen is illuminated with nearly circularly analyser. In the top right image of Fig. 2, the specimen is
polarized light and the traditional compensator is replaced illuminated with elliptically polarized light and astral rays
by two electro-optical modulators (Fig. 1). The two electro- orientated at 45" to the principal axes of the polarization
optical modulators are currently implemented as liquid ellipse appear either brighter (vertical microtubules) or
crystal devices that function as linear retarders whose darker (horizontal microtubules) than the background. The
retardance values are varied independently by a voltage horizontal/vertical position of dark and bright quadrants of
applied to each device. As indicated in Fig. 1, the slow axes the aster in the top right image of Fig. 2 corresponds to the
of the linear retarders are mutually orientated at 45". By horizontal/vertical orientation of the slow/fast axis of linear
adjusting the liquid crystal retardances, this combination of retarder B. On the other hand, the diagonal position of dark
variable retarders can compensate any specimen retar- and bright quadrants of the aster in the bottom two images
dance, regardless of orientation and magnitude, so that a of Fig. 2 corresponds to the diagonal orientation of the slow/
minimum of light equal to extinction passes through the fast axis of retarder A.
analyser. The two variable retarders thus act together as a We want to emphasize here that during recordings of all
universal compensator. four images in Fig. 2, no part of the microscope set-up,
To illustrate the workings of the universal compensator, including the specimen, was mechanically rotated. To
we show four different images (Fig. 2) of the same specimen, obtain the four images, only the voltages applied to the
a so-called aster formed in lysate prepared from eggs of the liquid crystals were changed. Therefore, all four images
surf clam (Oldenbourg et al., 1993). Asters are convenient recorded with the CCD camera are in perfect register, with
test objects because their radial symmetric structure corresponding picture elements representing intensities of
contains astral rays orientated in all directions, with each the same object element. Experimental and theoretical
ray possessing a linear retardance of a fraction of one results obtained by Hansen et al. (1988) indicate that the
nanometre to several nanometres and a slow axis diffraction anomaly reported for polarizing microscope
orientated parallel to the ray axis (astral rays are bundles systems using linearly polarized light (Inoue & Kubota.

(Q 1995 The Royal Microscopical Society. Journal of Microscopy. 180. 140-147

 Fig. 2. CCD-camera recorded images of a single
aster consisting of microtubules radiating in all
directions from the centrosome. a microtubule
organizing centre. Images of the aster were
recorded using four different settings of retarders
A and B with retardance settings indicated
above and below the images. X is 546nm, the
wavelength of the illuminating light. (Images
are contrast enhanced for better printing.)

1958) is substantially reduced in microscopes which use B = (X/2) + Pmin(the subscript min indicates that for these
nearly circularly polarized light, as is the case in the new retarder settings a minimum intensity equal to extinction is
pol-scope. Optimal image registration and absence of recorded for the given specimen retardance characterized by
diffraction anomaly is particularly important for measure- R and 0):
ments requiring high spatial resolution.
The retarder settings for images in Fig. 2 compensate the
R= JGz, (1)

retardance of astral rays orientated either horizontally (top

right image) or diagonally (bottom two images). We found
that astral rays of any other orientation can also be
compensated with retarder settings A and B that in general The Sign(amin)function is either +1 for positive amin or - 1
will both deviate from their extinction settings of a quarter for negative amin.
wave (A) and half wave (B). In fact, we have found an Equations (1) and (2) were first found by modelling the
expression that relates the retarder settings A and B to the optical set-up with a computer program written in Mathema-
specific specimen retardance that these retarder settings tics (Wolfram Research Inc., Champaign, IL). The program
compensate. In the following formulae the magnitude of simulates a single ray that first passes through the linear
specimen retardance is given as R, typically in nanometres, polarizer, then the two variable retarders, then the specimen
and its slow axis orientation is 0. 8 is measured in degrees and finally the circular analyser. The computer simulationwas
in the x-y plane starting from the positive x-axis (refer to based on the Stokes-formalism to describe the polarization
coordinate axes specified in Fig. 1).The retarder settings A state of the light ray. The polarization-sensitive devices were
and B that compensate the specimen retardance are given represented by their corresponding Mueller-matricesto predict
in terms of aminand p, with A = (X/4) + aminand the new polarization state after the ray passed through the

0 1995 The Royal Microscopical Society. lourml oJMicroscopy. 180. 140-147

Fig. 3. Images computed From raw image data shown in Fig. 2. In the Retardance image in the top centre, white corresponds to a maximum
retardance magnitude of 3.9 nm, black corresponds to zero retardance, with greys indicating retardance values in between. In the Azimuth
image on the top left, black indicates zero azimuth, i.e. slow axis of measured retardance orientated horizontally, medium grey indicates azi-
muth 90", i.e. slow axis vertical, and white indicates an azimuth of 18O0,which corresponds again to a horizontal slow axis. Other grey
values indicate azimuth angles between 0" and 180". In the right image, both Retardance and Azimuth are given, with the azimuth indi-
cated by lines with orientations between 0" and 180". To reduce the complexity of this image we chose to display the azimuth on regular grid
points (indtcated by small closed circles) at a much lower resolution than the original measurements permits.

device. The effect of several linked devices can be computed by involving the four images of Fig. 2. By way of explanation.
multiplying the corresponding matrices in the right order (see we consider the intensities II to l4 recorded in a single.
e.g. Shurcliff, 1962; Kliger et al., 1990). A more detailed corresponding picture element (pixel) of the four images.
description of the mathematical foundations of the algorithms Each of these four intensities depends on the compensator
used with the new pol-scope will be published separately. settings A = (XI4) + a and B = (XI2) + $ and on the
Using the computer simulation we found that a given retardance of the small specimen region that is imaged at
specimen retardance of any magnitude and azimuth can be this pixel. When the specimen retardance and the changes
compensated with the two variable retarders, so that a in compensator settings ( a and p) are small in retardance
minimum intensity equal to extinction is recorded with --
magnitude (< XI20 25 nm) the following quadratic
particular retarder settings. Equations (1) and (2), however, expression holds:
only apply if the specimen retardance is small. For a
specimen retardance equal to X/20, with X the wavelength
of light used in the measurement, the systematic error
introduced by using Eqs. (1) and (2) to compute the
specimen retardance from the measured (a,,,, Pmin)is with n from 1 to 4 for each pixel. (Eq. 3 is a quadratic
about 2%. With a specimen retardance of XI40 the error in approximation of the complete expression involving the
the computed magnitude and azimuth reduces to about square of trigonometric functions of retardances. The
0.5%. In fact, for measuring small birefringence values, we approximation introduces a systematic error which is less
have identified a very efficient measurement process and than 3% for retardances less than 2 5 nm.) In the first, top
computational algorithm, which we discuss next. left image of Fig. 2, for example. a , and PI were both zero
and the intensity II is proportional to the square of the
magnitude of specimen retardance R. lo is the proportion-
ality factor and is equal to the intensity observed at total
3. Theory of measuring small, linear retardances
transmission. Imi, is the intensity observed when the
(< X/20) specimen retardance is compensated by the variable
The four images shown in Fig. 2 are indeed sufficient to retarder settings, i.e. when a,,= amin and /?,] = Pmin.Imin
compute the specimen retardance and orientation of the is the intensity equal to extinction and can be non-zero due
slow axis (azimuth) in every image point. In other words, to imperfections in the optics, or due to multiple scattering
the retardance of every astral ray, regardless of its in the specimen.
orientation, can be computed using simple image arithmetic Images in Fig. 2 were recorded using four Exed retarder

1995 The Royal Microscopical Society. Jol~rnaloJMicrosropg. 180. 140-147

144 R . O L D E N B O U R G ~ l rG . M E I

settings: projected into the specimen space and cause an apparent

retardance ('background retardance') that varies slowly
(a, = 0,[31 = O), (4a) across the viewing field (Fig. 4). In Fig. 4 we show images
recorded from a preparation of single mitrotubules in
aqueous solution (Tran rt a!.. 1994). As can be seen.
images of these weakly birefringent objects are seriously
(a4= +x,/&= o), (44 affected by background retardances. While the background
azimuth generally can take on any angle (see Fig. 41, we
with X = 1 6 nm retardance. Four independent settings are found that with care in the selection of strain-free
necessary because Eq. (3) has four unknown parameters microscope optics, slide and cover glass, the magnitude of
a m i nPmin)
(I0,Iminr , and three known parameters (I,,,a,,and background retardance can be quite small and uniform over
P,,) With the particular settings chosen in Eq. (4). the four a significant area, in the range of 0-0.5 nm. Furthermore.
quadratic equations derived from Eq. (3) have simple we found that for a given pixel, the background retardance
solutions for the unknowns amin and Pmin: is nearly constant in time and can be subtracted from the
specimen retardance using the following procedure.
First, four specimen images are taken of the kind
shown in Fig. 2. Then, the specimen, which is sandwiched
between microscope slide and cover glass, is moved
sideways on the microscope stage until a clear area with
Note that a,,, and [j, are calculated independent of lo no specimen birefringence is imaged. A second set of
and I,,,. Therefore, these parameters do not need to be four images is recorded with the same voltage settings as
determined separately. To find the desired expression for the used before. The second set of four images measure the
specimen retardance R and azimuth O in terms of the background retardances. Next, the ~ , 1 1 , , and
1 ~ ~ /J,,lnBG values
measured intensities l1 to 14, we simply need to replace in of the background retardances are computed using
Eqs. (1)and (2) the terms cr,,, and ,Om,, with the right sides Eqs. (5) and (6) and the intensities recorded in the
of Eqs. ( 5 ) and (6). In fact, the computation of specimen background set of images. Similarly, with the tirst set of
retardance can be done for the entire viewing field at once images, representing specimen plus background retar-
by implementing Eqs. (1)and (2) with (5) and (6) as two dances, the amlnsp, ,,
BC and ,L?m,,s,,, values are computed.
sequences of image arithmetic functions. The first sequence To compute the retardance magnitude and azimuth
produces an image representing the magnitude of specimen associated with the specimen only, a,,, and in Eqs.
retardance, the second sequence produces an image (1) and (2) are substituted with
representing the azimuth (Fig. 3). The images can be
displayed either separately or combined using some special
and
coding scheme. Thus, displayed images represent a direct
visualization of specimen retardance measured at the spatial
resolution of the light microscope optics. Furthermore, since The images shown in Fig. 4 demonstrate the effectiveness of
the computed specimen retardance is independent of 1" this background correction procedure.
and I,,,, the resultant images are corrected for shading The above procedure can also be used to measure small
introduced by inhomogeneous specimen illumination, and changes in retardance over time or between different
for unpolarized background light, which stems, for example, focus levels. Figure 5 documents the retardance of a
from non-ideal polarizers, or from excessive light scattering magneto-optical disc substrate, which was measured by
by turbid samples. first focusing on the disc surface imaging the moulded pits
and grooves on the disc surface. The subsequent back-
ground set of images was taken by focusing a few
4. Background correction procedure
micrometres above the disc surface, completely blurring
With the analysis outlined above the instrument is capable of the surface features. The retardance images shown in Fig. 5
measuring specimen retardances between approximately 0.5 were then computed by using the in-focus set of images as
and 2 5 nm. The upper limit is given by the approximations specimen set and the out-of-focus set as background.
used in deriving Eqs. (1)-(3). while the lower limit is caused by Figure 5 demonstrates the result of yet a different kind of
polarized background light that stems from polarization background subtraction that is achieved using the liquid
aberrations introduced, in part, by the strong curvature of crystal retarders. The magneto-optical disc substrate
high-NA lens surfaces in the optical train (Inoue & Hyde, possesses a uniform retardance of about l 8 n m in
1 957; Chipman, 1989). Polarization aberrations originat- magnitude (fast axis orientated radially from the disc
ing from different parts in the microscope optical train are centre) owing to the partial alignment of polymers during

(y; 1995 The Royal Microscopical Soclety. Juurnul uj M~cron.op!j. 180. 140-1-17
 NEW P O L A R I Z E D L I G H T MICKOSC'OPE 145

Fig. 4. Demonstration of background correction procedure on retardance and azimuth images of microtubules (single microtubules are
2 5 nm thick, several micrometres long, stiff and weakly birefringent biopolymers). In the sample imaged here microtubules grow off
the ends of a short piece of axoneme, a birefringent rod-shaped polymerization seed (near image centre) which is stuck to the microscope
coverglass. The specimen images in the top row (left: retardance image, right: azimuth image) are noticeably affected by background
retardance. The centre row shows images of the background alone recorded after the preparation was moved sideways to vlew a clear speci-
men area. The bottom row shows the same specimen images as in the top row after the background correction procedure was appl~ed(see
text). The background-corrected specimen images show a uniform background with noise due to uncertainties in measured intensities (see
text). The retardance of microtubule bundles, containing one or more microtubules growing off the ends of axonemes. was measured to be in
the range of 0.07-0.21 nm. (Sample preparation and images by Phong T. Tran.)

the extrusion process. The uniform retardance is not visible After background retardances are properly corrected for,
in the images of Fig. 5 because it was optically compensated the minimum specimen retardance that can be measured
during the measurement by readjusting the liquid crystal depends on the uncertainties of recorded image intensities.
retarder settings. Therefore, any retardance, due to grooves In Fig. 4 for example, the bottom images, which are
and pits for example, leading to deviations from the uniform background corrected, show random fluctuations in
background retardance was measured very sensitively. measured retardance due to uncertainties in measured
The described optical method of compensating back- image intensities. In CCI) cameras, uncertainties in
ground retardance is most useful to subtract a uniform bias measured image intensities are due to several sources.
retardation in the sample or anywhere else in the optical such as photon statistics noise, CCD read-out noise and
train of the microscope. However, variations of background amplifier noise (in our measurements, light levels were
retardance within a given field of view cannot be high enough that photon statistics noise was negligible).
compensated using the optical method, but these local In image parts with no specimen retardance. random
retardance variations are corrected with the computational intensity fluctuations result in an apparent retardance
background correction procedure described earlier. The in each pixel that has a root-mean-square magnitude
optical and computational methods of background correc- proportional to the standard deviation of the fluctuations
tion are therefore complementary. and an azimuth angle that varies randomly between 0" and

:?I 1995 The Royal Microscopical Society, Inurnel 01Mitrostnpg. 1 8 0 140-1 47

 (Technical Video, Woods Hole, MA) and narrow band pass
interference filter (546 nm, 1 0 n m FWHM, Omega, Brat-
tleboro, VT). Linear polarizer and left circular analyser
(Meadowlark Optics. Longmont, CO) are inserted into the
microscope optical path at the same places where they are
located for traditional polarized light microscopy. The liquid
crystal variable retarders (Cambridge Research and Instru-
mentation Inc. (CRI),Cambridge. MA) are placed in custom-
made rotatable mounts on top of the field lens near the base
of the microscope stand. The magnitude of retardance of the
liquid crystal devices is adjusted with a n AC square wave
controlled through special electronic circuitry (CRI, Cam-
bridge, MA). Liquid crystal retarders were calibrated using
a n optical bench set-up, including a monochromatic light
source, high extinction polarizers and a Babinet Soleil
compensator. The liquid crystals in the microscope set-up
are controlled by a desk top computer (Quadra 800. Apple
Computer, Cupertino, CA) used for instrument control and
image processing. A scientific grade video camera (CCD
C72, Dage-MTI, Michigan City, IN) records images of
specimens illuminated under different polarization con-
ditions (Fig. 2). A frame grabber board (LG-3, Scion Corp..
Frederick, MD) in the computer digitizes the video signal
( 6 4 0 x 4 8 0 pixels, 8 bit per pixel) and stores up to 128
individual frames captured at video rate ( 3 0 frames per
second). In addition to digitizing the video signal, the LG-3
frame grabber board also provides the control voltages for
the liquid crystal devices, with the capability of synchroniz-
ing frame capture with changes in liquid crystal settings.
The software for instrument control and image processing is
based on NIH Image, a public domain program by Wayne
Fig. 5. Surface of magneto-optic disc substrate (polycarbonate)
Rasband (National Institutes of Health, Bethesda, MD; NIH
imaged with the new pol-scope. Top image: retardance magnitude
Image is available by anonymous ftp from zippy.nimh.nih.
near moulded pits and grooves (maximum retardance 5,6nm,
grooves are 1 . 3 pm apart). As explained in the text, the displayed gov/pub/nih-image). We wrote specialized NIH Image
retardances are the values measured as deviation from the uniform macros, procedures and source code to semi-automate the
retardance (approximately 18 nm, slow axis parallel groove direc- task of image capture, retarder control, data processing,
tion) of the polycarbonate bulk material. Lower image: small boxed computation and display.
area in top panel enlarged with measured azimuth orientations Before being digitized, images are analog enhanced in the
indicated for each pixel by short lines. camera controller box using the camera gain and black
level controls. Manual controls are first set without a
180". With the instrumentation described in the next section birefringent specimen in the optical path (typically, the
we have achieved noise levels that correspond to 0.1 nm root- specimen, sandwiched between slide and coverglass, is
mean-square retardance per pixel with proportionately moved sideways until a clear region is imaged). The camera
reduced noise when averaged over several pixels. black level is adjusted with the extinction setting (A = X/4
and B = X/2) to render the background image nearly
black. Then the camera gain is adjusted to make image
5. Instrumentation and procedures
intensities with any of the other three retarder settings
The new pol-scope system illustrated in Fig. 1 was 6rst read a medium grey (typically 1 5 0 out of 2 5 5 intensity
implemented on a n upright research grade microscope values). With the camera controls set, the liquid crystal
(Microphot-SA. Nikon Inc., Melville, NY; images in Figs. 2- control voltages are now fine adjusted so as to give
5 were recorded using a 6 0 x 11.4-NA Plan Apochromatic the lowest intensity for the extinction setting and an equal,
objective lens and 1.4-NA apochromatic condenser). For medium grey intensity for all three other settings (see
intense monochromatic light and homogeneous illumina- Eqs. 4a-4d). This procedure of adjusting the retarder
tion we use a mercury arc lamp with Ellis light scrambler voltages to read equal image intensities for settings given

(r)1995 The Royal Microscopical Society, lourr~al(I] Microwop!,. 180. 140-147

by Eqs. (4b), (4c) and (4d) ensures that the retardance provided by Shinya Inoue. We thank our collaborators
increment X is the same for all three settings (X is typically Robert E. Palazzo, University of Kansas. Edward D. Salmon
16 nm, see Fig. 2). and Phong T. Tran, University of North Carolina, for helpful
After all the necessary adjustments are completed, a stack discussions and for generously providing samples. We wish
of background images at the four retarder settings are to thank Robert A. Knudson for assisting in the design and
captured. To reduce noise typically 8 to 32 frames are fabrication of mechano-optical parts. We also thank Nikon
recorded for each retarder setting and frame averaged. The Corporation for instrument support. The instrument devel-
stack of four, frame averaged images. one for each retarder opment is supported by the National Institutes of Health
setting, is then used to compute two images representing grants R01 GM49210 awarded to R.O. and R37 GM31617
the ( Y , , , ~ G and PmlnnGbackground values in each pixel (see awarded to S.I.
section on background correction procedure). Then the
specimen of interest is moved into the viewing field and
References
another stack of four images is recorded (see Fig. 2).
With these four images the specimen retardance is Allen. R.D., Brault. 1. & Moore. K.1). (1963) A new method of
computed (see Fig. 3). For best results the background polarization microscopic analysis I. Scanning with a birefrin-
correction procedure discussed earlier is applied during the gence detection system. 1. C[dl Biol. 18, 22 3-2 3 5.
calculation. Chamot. E.M. & Mason. C.W. (1958) Hrirltlbook qf Cl~t~rrric~~l
Microscopy, 2nd edn. John Wiley & Sons. New York.
With the instrumentation and procedures described
Chipman, R.A. ( 1989) Polarization analysis of optical systems. Opt.
above we achieved a noise level corresponding to 0.1 nm
Eng. 28, 90-99.
root mean square retardance with single frame capture of Hansen. E.W.. Conchello, J.A. & Allen. R.1). (1988) Restoring
the specimen stack (no frame averaging). The noise level is image quality in the polarizing microscope: analysis of the Allen
a direct measure of the sensitivity of the instrument. video-enhanced contrast method. 1. Opt. Sot. Anr. 5 , 1836-
Specimen retardances smaller than the noise level remain 1847.
undetected, while specimen retardances larger than the Hartshorne, N.H. & Stuart, A. (1960) Crystrrls lir~dthc' Polctrisirr~g
noise floor clearly stand out in the image. With single frame Microscope,: A Handbook Jor Clrcvnists rrrrd Othc~rs.3rd edn. Arnold.
capture, the elapsed time between the first frame captured London.
with the first retarder setting and the fourth frame captured InouC. S. (1986) Video Microscol~y.Plenum Press. New York.
with the fourth retarder setting is 0.43 s or 1 3 video-frame Inoue. S. & Hyde. W.L. (1957) Studies on depolarization of light
at microscope lens surfaces 11. The simultaneous realization of
times: one frame for each capture and three frames waiting
high resolution and high sensitivity with the polarizing
time for each change of the liquid crystal retarder settings
microscope. 1. Biophys. Biochenl. Cgtol. 3 , 831-8 38.
(the liquid crystal devices in use settle to a new retardance InouC, S. & Kubota. H. (1958) Diffraction anomaly in polarizing
value in just under 100ms or three frame times). With a microscopes. Nature. 182, 1725- 1726.
static or nearly static specimen, several video frames can be Inoue. S. & Sato. H. (1966) Deoxyribonucleic acid arrangement in
averaged for each retarder setting, resulting in a reduction living sperm. MoIecrrlar Architrctrtrr in Cc41 Pftgsiologg (ed. by T .
of camera noise and an increase in sensitivity. With slowly Hayashi and A. G. Szent-Gyorgi). pp. 209-298. Prentice-Hall.
moving specimens, frame averaging can increase sensitivity Englewood Cliffs. NJ.
at the expense of time and spatial resolution. Kliger, D.S., Lewis. J.W. & Randall, C.E. (1990) Polarized Light irl
Optics and Spectroscopy. Academic Press, New York.
McCrone. W.C. (1991) Light microscopy. Pl~!ysirrtl Mrtholb of
6. Conclusions Chemistry (ed. by B. W. Rossiter and J. F. Hamilton), pp. 343-
443. John Wiley & Sons. Inc.
The new instrument dramatically enhances the unique
Mei. G. & Oldenbourg. R. (1994) Fast imaging polarimetry with
capabilities of the traditional polarized light microscope precision universal compensator. Proceedings oJ the' Polurizatior~
to measure submicroscopic molecular order. It can do Analysis and Measurmlmt 11, pp. 29-39. SPIE-The Inter-
so independent of specimen orientation at high spatial national Society for Optical Engineering.
and temporal resolution, concurrently for the whole Oldenbourg, R., Mei, G. & Palazzo, R.E. (199 3) Asters in lysates of
field of view. These significant advances will launch Spisula oocytes observed with a new type of polarized light
new investigations into the structure and dynamics of microscope. Biol. Bull. 185, 288.
molecular organization in application areas that tradition- Shurcliff, W.A. (1962) Polarized Light. Prod~cctiorland Usr. Harvard
ally harness the analytical power of the polarized light University Press, Cambridge, MA.
microscope. Tran, P.T., Inoue, S., Salmon, E.D. & Oldenbourg, R. (1994) Muscle
fine structure and microtubule birefringence measured with a
new pol-scope. Biol. Bull. 187. 244-245.
Acknowledgments
We gratefully acknowledge the inspiration and support

ri? 199 5 The Royal Microscopic;rl Society. Iournol IIJ Mirrosropy. 180, 140- 147