Histochemistry

Histochemistry (1987) 86:331-336

9 Springer-Verlag 1987

Understanding Romanowsky staining

I: The Romanowsky-Giemsa effect in blood smears
R.W. Horobin* and K.J. Walter**

Department of Anatomy and Cell Biology, The University, Sheffield S10 2TN, England
Accepted August 2, 1986
Summary. Normal blood smears were stained by the standardised azure B-eosin Y Romanowsky procedure recently
introduced by the ICSH, and the classical picture resulted.
The effects of varying the times and temperature of staining,
the composition of the solvent (buffer concentration, methanol content, & pH), the concentration of the dyes, and
the mode of fixation were studied. The results are best understood in terms of the following staining mechanism. Initial colouration involves simple acid and basic dyeing. Eosin
yields red erythrocytes and eosinophil granules. Azure B
very rapidly gives rise to blue stained chromatin, neutrophil
specific granules, platelets and ribosome-rich cytoplasms;
also to violet basophil granules. Subsequently the azure B
in certain structures combines with eosin to give purple
azure B-eosin complexes, leaving other structures with their
initial colours. The selectivity of complex formation is controlled by rate of entry of eosin into azure B stained structures. Only faster staining structures (i.e. chromatin, neutrophil specific granules, and platelets) permit formation of
the purple complex in the standard method. This staining
mechanism illuminates scientific problems (e.g. the nature
of 'toxic' granules) and assists technical trouble-shooting
(e.g. why nuclei sometimes stain blue, not purple).

Introduction

At their simplest Romanowsky stains contain the red acid

dye eosin Y plus the blue basic dye azure B, although additional methylene blue homologues are often present (Wittekind 1983). Since the nineteenth century Romanowsky
stains such as Giemsa's and Leishman's have provided the
basis of recognising, and indeed categorising, the normal
and pathological cells of peripheral blood and bone marrow.
An advantage of Romanowsky staining for routine haematology is the highly polychrome image produced; red
erythrocytes and eosinophil granules; blue lymphocyte cytoplasms; intensely purple nuclear chromatin, azurophil
and neutrophil granules, and platelets. Formation of the
purple colour, i.e. the Romanowsky-Giemsa effect (RGE),
is peculiar to these stains. Compare the Romanowsky staining pattern with the colours produced by another standard
* To whom offprint requests should be sent
** Current address: Fox Hill School, Bracknell RG12 4AY, England

red acid dye-blue basic dye pair, hematoxylin and eosin

(H & E). With comparably diluted eosin H & E also gives
rise to red stained erythrocytes and to purplish-blue chromatin. Lymphocyte cytoplasms however also stain purplish-blue; whilst eosinophil and neutrophil granules stain
the same colour as the rest of their cytoplasms. So distinguishing between nuclear chromatin and basophilic cytoplasms, and between granules and cytoplasm, is harder with
H & E than with Romanowsky staining.
In spite of their advantageous staining qualities and long
history, Romanowsky stains remain mechanistically obscure. The bases of the polychromasia, in particular the
RGE, and of the influences of practical variables such as
dye structure, solvent composition and pH, and of staining
time all remain enigmatic; see for instance the 'Conclusions' of Bentley and his collaborators (1979). This lack
of understanding has limited interpretation of staining results, and has made technical trouble-shooting a matter
of trial and error.
Difficulties of understanding are in large part a consequence of traditional modes of preparing Romanowsky
stains. Chromatography shows these methods give variable
and polychromatic dye mixtures (Marshall 1978). So reliable mechanistic studies are limited to those using reliable
analytical methods; namely the investigations of Wittekind
and his group (Wittekind 1983), Marshall and collaborators
(Marshall 1980), and Lillie (cited in Clark and Kasten 1983,
p. 19).
The work described here used the Romanowsky stain
and staining procedure recently recommended by a Working Party of the International Committee for Standardisation in Haematology (ICSH 1984). The stain contained only
azure B and eosin Y. We also systematically varied staining
time; dye concentrations; solvent pH; and water-organic
solvent ratio. The results were interpreted in terms of the
most parsimonious of theories. Truth is doubtless more
complex than our interpretations imply, but we consider
that we have both isolated and cmphasised the most crucial
mechanistic factors in Romanowsky staining.
Materials and methods

Blood smears were prepared manually in the usual way. Blood

was collected into heparinised capillary tubes, and a large number
of smears prepared from each sample. These smears were air dried,
fixed in methanol for 5 rain, washed in distilled water for t min,
and again air dried.

332
The "Standard-Romanowsky-Giemsa-Stamml6sung" stock
solution (Heyl Chemisch-Pharmazeutische Fabrik, Berlin) had been
produced according to the recent ICSH recommendations by disolving 3 g of azure B perchlorate in 400 ml of dimethyl sulphoxide
at 37~ C, disolving 1 g of eosin Y in 600 ml of methanol, and
mixing the two dye solutions. Stock solution purity was checked
using reverse phase thin layer chromatography (Proctor and Horobin 1986), and found to be satisfactory. Dye standards used for
this were prepared in the laboratory of Prof. D. Wittekind, Freiburg.
The standard working solution was prepared by diluting stock
solution with 0.05 M pH 6.8 HEPES buffer to which had been
added 50 ml/1 dimethylsulphoxide. The volume ratio of buffer to
stock solution was 15:1. The buffer was prepared from N-2-hydroxyethylpiperazine-N1-2-ethanesulphonic acid (Sigma Chemical
Co.) by the procedure of Perrin and Dempsey (1974). Variations
on this standard solution are outlined below.
The standard staining procedure was that recommended by the
ICSH (1984). Fixed smears were stained in a Coplin jar with working solution for 25 min; then rinsed in distilled water for 1 min,
and air dried. Once dry the smears were cleared with Histoclear
(National Diagnostic (UK) Ltd.); and mounted in Polymount, a
polystyrene medium. Variations on this standard procedure are
outlined below.
Staining times were varied, with all other aspects of the standard method remaining the same. Times used were 30, 60 and
90 s; 5, 10, 15, 20 and 25 rain; 1, 1.5, 2, 4, 12, 17, 24, and 28 h.
For the longer staining times the staining solutions were changed
every 4 h, to avoid significant dye precipitation.
Staining solutions with reduced dye concentration (both a half
and a tenth of standard) were prepared, using suitable dilutions
with buffer. Smears were stained for 30 s-5 rain.
Stock solutions were prepared using single dyes, both azure
B and eosin Y. These, and the working solutions derived from
them, were prepared in the ICSH-recommended manner, but ommitring one of the dyes. Smears were stained in the standard manner, and also with staining times varied over the range 30 s-72 h.
The pH of the standard working solution was varied by diluting
the stock solution with buffers of different pHs. A 0.0286 M universal buffer (Perrin and Dempsey 1974) was used to obtained working
solutions of pH 4, 6.8 and 8. This was used with staining times
of 25 min, and i and 2 h. A 0.01 M HEPES buffer was used at
pH 6.6, 7.5 and 8; with other variables kept standard.
A high ionic strength working solution was made up with
0.05 M buffer, c.f. the standard 0.01 M. Smears were stained for
the standard time immediately after preparing the working stain,
and also after allowing it to stand on the bench for 2 and 4.5 h.
Working solutions containing extra organic solvent were investigated. The water used to prepare the working solution was partially replaced by varying concentrations of aqueous methanol,
100%, 50%, 25% and 10% by volume. Smears were stained for
25 rain at all methanol concentrations, and also for 2.5 h in the
solution using 25% methanol.
The effects of alternative fixatives were compared. One batch
of smears was fixed in formaldehyde vapour, by standing the slides
in a staining dish above a layer of 40% w/v aqueous formaldehyde.
The system was stood for an hour before inserting the smears.
Fixation time, and subsequent staining times, were varied over
the range 1-30 rain. Another set of smears were fixed in this way
(for 30 and 40 rain), then postfixed in methanol for 5 rain. A third
set was fixed in the standard manner with methanol, then postfixed
for 30 and 40 min in formaldehyde vapour.
Using the standard working stain, the effects of staining temperature were investigated. Coplin jars containing stain were placed
in a water bath, and after equilibration to the desired temperature,
staining was carried out: at 10~ C for 25 rain; at 30~ C for 25 min;
at 35~ C for 5 and 25 min; and at 45 ~ C for 5 min.
Asay of stain in solution was carried out by spectrophotometry.
The stain was filtered and the filtrate diluted with aqueous methanol to yield a final methanol-water volume ratio of 1 : 1. The degree
of dilution was adjusted to provide a maximum absorbance of

ca. 1. The absorbtion spectra of each stain was measured between

450-750 nm using a Varian DMS 90 uv/visible spectrophotometer.
Staining intensities of smears were classified independantly by
two observors on a 1-10 scale ranging from 'very weak staining'
to 'intense staining'; 'no staining' was classified as 0. Objects for
which intensity data is presented were observed at least 10 times
on each slide; the information given constitutes the average impression.

Results and discussion

An explanation for "'the colour purple "" (i.e. the R G E )

Smears prepared using the I C S H standard stain and staining conditions showed the classical staining pattern. Erythrocytes and eosinophil granules were red; lymphocyte cytoplasms blue; m o n o c y t e cytoplasms grey-blue; basophil
granules violet-purple; azurophil and neutrophil specific
granules, chromatin, and platelets were purple.
The basis of certain o f these colours is straight forward.
U p t a k e o f the blue basic dye (azure B) by R N A - r i c h cytoplasms will u n d o u b t e d l y be o r t h o c h r o m a t i c basic dyeing,
whilst the violet staining of heparin-containing basophil
granules by azure B will be clue to m e t a c h r o m a s i a ; u p t a k e
o f the red acid dye (eosin Y) by erythrocytes and eosinophil
granules will be acid dyeing (Baker 1958, chap 10). These
interpretations are validated by the congruence of observed
p H effects with the effects expected on theoretical grounds
for simple acid and basic dyeing. F o r instance obvious basophilia o f erythrocytes developed on increasing the pH,
whilst acidophilia of erythrocytes rose with a fall in pH.
However the colour purple seen in chromatin, azurophil
and neutrophil specific granules, and platelets is generally
regarded as a puzzle (Marshall 1980; W i t t e k i n d 1983).
Some workers have suggested that eosin plays no p a r t in
the R G E , and that the colour purple is merely naetachromasia (Baker 1958, p 269; Commings 1975). Others however
found that eosin was essential for the R G E to occur, and
suggested that the colour purple involved azure B-eosin
Y interactions. These, they argued, t o o k place only in the
presense of suitable macromolecules. These can apparently
be varied in character, e.g. D N A (Sumner 1980), D N A protein complexes (Zanker 1981), protein-mucopolysaccharide complexes (Wittekind and Kretschmer 1980), and gelatin (Wittekind 1983).
In the present paper we demonstrate the value of turning
this latter view o f R o m a n o w s k y staining on its head. In
our view it is not the colour purple which needs explanation,
but the colours red and blue.
W h e n R o m a n o w s k y staining solutions are allowed to
stand, precipitation o f a purple complex (an azure eosinate)
commences, and this continues until no eosin remains in
solution. Hence we propose that purple staining o f granules
and chromatin is merely the f o r m a t i o n o f an azure-eosin
complex, inside the cells instead of in solution. This hypothesis parallels the views of P a p p e n h e i m and Nocht, working in the first decade of this century (cited by Baker 1958,
p 272), and also the arguments developed for c h r o m o s o m e
b a n d i n g by Sumner (1980).
W h e t h e r the purple, intracellular azure-eosin complex
is in fact identical to the azure eosinate precipitated from
solution is uncertain; certainly Wittekind and his associates
would dispute it. The present account puts such questions
to on side. Here we focus on why the intracellular azure-

333
Table 1. Rates of formation of "the colour purpIe" during Romanowsky staining
Structure stained

Erythrocytes
Lymphocyte azurophil granules
Lymphocyte chromatin
Lymphocyte cytoplasm
Neutrophil chromatin
NeutrophiI specific granules
Platelets

Observed colour after staining for stated time

30 s

90 s

25 min

28 h

faint pink
very pale purple
faint blue
very pale blue
faint blue
no staining
very pale blue

pink
faint purple
blue
faint blue
blue
faint blue
blue

red
purple
purple
blue
purple
purple
purple

purple
intense purple
intense purple
purple
intense purple
intense purple
intense purple

eosin complex is formed in some structures but not in

others.
Why do all stained cellular components not become purple? In fact they do, though only when the period of staining is sufficiently extended: see Table I for selected results,
covering staining times from 30 s to 30 h. An increasingly
purple staining with extended staining times has been previously reported, both numerically (Bentley et al. 1979,
Fig. 5) and pictorially (Wittekind 1983, Fig. 9a). However
no mechanistic significance was attributed to the phenomenon.
The causes o f the poIychromasia - a proposed R G E staining
mechanism

As previously reported by others (Lillie 1943; Sumner 1980;

Wittekind 1979), we found that initial dye uptake did not
give the complete polychrome effect, but instead gave blue
stained chromatin, lymphocyte cytoplasms and neutrophil
specific granules; purple stained azurophil and basophil
granules; and pink stained erythrocytes and eosinophil
granules. All initially blue or red sites eventually turned
purple, some very much faster than others: as seen in Table 1.
Polychromasia thus largely depended on a rapid initial
acid and basic dyeing, which was usually orthochromatic,
though metachromatic in basophil granules; followed by
formation of a purple azure-eosin complex. This latter process occured most rapidly in azurophil granules; nearly as
rapidly in chromatin, neutrophil specific granules and plate
lets; slowly in RNA-rich lymphocyte cytoplasms and
eosinophil granules; and exceedingly slowly in erythrocytes.
Though azurophil granules are purple from their first appearance, this is not metachromasia. It does no~ arise in
the absense of eosin, unlike the violet-purple of the basophil
granules.
So the selective purple staining is a rate of reaction phenomenon. We must now enquire what factor controls the
rate of formation of the purple product. It is probably the
differential rates of entry of the second dye into the various
structures already stained by the first dye. Quantitative studies of dye penetration into fixed tissue sections showed
that nuclear chromatin stained faster than RNA-rich cytoplasms (Goldstein 1963), and indicated that erythrocytes
were extremely slow staining (Horobin 1974). Such conclusions correlate with the rapid appearance of the colour purple in chromatin in Romanowsky staining, compared to
the slow appearance in RNA-rich cytoplasms; with erythrocytes being slower still.

Formalin fixed tissue sections could of course behave

differently to methanol fixed cell smears. However in blood
smears stained with azure B alone, azurophil granules
stained most rapidly, followed by chromatin, with RNArich cytoplasms being slower still. So the relative staining
rates of similar structures are probably also similar in cell
smears and tissue sections.
Support for the proposal that the colour purple corresponds to formation of an azure-eosin complex, with this
'intracellular precipitation' being controlled by the rate of
entry of the second dye, comes from another study by Goldstein (1965). He demonstrated that, in formalin fixed tissue
sections, rapid staining occured in structures of low refractive index, and slow staining in structures of high refractive
index. The fast staining structures were in fact of low density and high porosity, and vice versa. Goldstein found that
the entities with the lowest refractive inex were glycogen
and mucin particles. Chromatin had a higher value, with
that of RNA-rich cytoplasms being higher still. The highest
values of all were those of eosinophil granules and erythrocytes. Since azurophil and neutrophil specific granules and
platelets are, like mucin granules, rich in glycosaminoglycans (Hardin and Spicer 1971; Ferdorko and Morse 1965;
Elliott and Knight 1975), this sequence parallels the rate
of formation of the colour purple.
Further evidence validating our proposed RGE staining
mechanism concerns monocyte lysosomes. These, unlike
those of neutrophils, do not become purple even after extended staining times. This is probably because monocyte
lysosomes, unlike those of neutrophils, lack the anionic glycosaminoglycan hyaluronic acid (Fedorko and Morse
1965), and so will not have a high affinity for the cationic
dye azure B.
Finally, the staining of neutrophil specific granules requires special mention. In neutrophils of normal individuals
these granules are not seen after very short staining times.
As staining times are increased the granules first become
blue, then purple, and eventually came to resemble 'toxic'
granules. The fact that blue stained chromatin may be seen
before blue stained specific granules is probably due to the
small size of the latter: human specific granules average
0.13 ~tm in diameter (Spitznagel et al. 1974). Deposition of
sufficient purple azure-esoin complex would increase granule diameter. If the purple complex is an azure eosinate
its precipitation will not be an equilibrium process. Formation of an azure eosinate would release anionic binding
sites on nucleic acids and glycosaminoglycans, which could
then bind more azure B (c.f. Sumner 1980). Such a continuous growth of precipitate would explain the 'toxic' appear-

334
ance following extended periods of staining. Non-artifactual toxic granules are in fact azurophil, not specific, granules; and have average diameters of 0.3 pm (McCall et al.
1969; Spitznagel et al. 1974). The appearance of neutrophils
could therefore vary (from 'agranular' to ' n o r m a l ' to
'toxic') as the pH or the azure B concentration goes from
low to high: a significant technical artifact.

Influence of practical staining variables

We have now seen that a R G E mechanism involving acid
and basic dyeing followed by a rate controlled 'precipitation' can explain the influence of staining time. We can
ask if this model also explains the influences of the other
practical variables.
Using the standard ICSH system, but reducing the dye
concentration, resulted in pale blue staining of structures
usually coloured purple, i.e. azurophil granules, chromatin
and platelets; with neutrophil specific granules being unstained. If staining times were extended beyond the standard 25 rain, then eventually purple staining did occur. The
more dilute the stain the paler the blue colouration at
25 min; and the longer the time needed for the colour purple to appear. Reduction in dye concentration reduced uptake of the leading dye, and this in turn reduced formation
of the azure-eosin complex.
When the buffer concentration was increased from
0.01 M to 0.05 M precipitation of dye from the staining
solution took place, resulting in pale but normal coloured
staining. When such solutions were stood on the bench
for a few hours before use, purple colours were not produced; chromatin and azurophil granules were pale blue,
and neutrophil granules were unstained. That this was a
failure to form an azure-eosin complex was indicated by
the visible spectra of the staining solution. These showed
that after a few hours of dye precipitation almost all the
eosin had been lost from the solution, though azure B was
still present.
Lowering the pH of the staining solution from neutrality
produced blue, not purple, chromatin and azurophil granules; unstained neutrophil specific granules (c.f. Bessis
1973); and more intense red staining of eosinophil granules
and erythrocytes. On raising the pH the colour purple was
slightly intensified (c.f. Wittekind 1979). The only obvious
effect of varying the chemical type of buffer used was that
erythrocytes were slightly bluer with a universal buffer than
with HEPES buffer of the same pH. Overall, pH effects
were in keeping with the assumption that the final colour
required uptake of the leading dye by acid or basic dyeing.
Thus reduction of azure B uptake onto nucleic acids as
the pH falls (c.f. Horn and Spicer 1964) results in a reduction in the amount of azure-eosin complex subsequently
formed. Elimination of staining of neutrophil specific granules at low pH is doubtless due to their basophilic character
depending on hyaluronate (Fedorko and Morse 1965). This
carboxylated polymer looses its anionic character at a higher pH than phosphated nucleic acids of chromatin and ribosomes, or the sulphated glycosaminoglycans of the azurophil granules (Hardin and Spicer 1971). Note that precipitation of azure eosinate from aqueous solutions is not markedly pH sensitive; complete precipitation of azure eosinate
occured from aqueous solutions over the pH range 4-10.
The consequences of increasing the methanol concentration in the staining solution were dramatic: the colour pur-

ple did not arise in the cells. Chromatin and azurophil granules stained blue, and neutrophil specific granules failed
to stain at all (c.f. Lillie 1944; Sumner 1980). However
at least with those solutions whose methanol concentrations
were only slightly raised, purple colouration did occur if
staining times were sufficiently extended. The initial blue
colour was only slightly paler than is seen with the standard
solution, so the methanol did not significantly inhibit the
uptake of the leading dye. Azure eosinates are methanol
soluble so the organic solvent could have inhibited formation of azure eosinate. In keeping with this, spectrophometry showed that precipitation of azure eosinate from
staining solutions fell as the methanol content of the solvent
rose.
Finally consider fixation. Formaldehyde and a simple
alcohol such as methanol will have quite different chemical
and physical effects on cells (e.g. Baker 1958, Part I; Horobin 1982, chap 3). Due to chemical modifications of cellular
proteins, formaldehyde inhibits uptake of acid dyes such
as eosin, and enhances uptake of basic dyes such as azure
B. Methanol has no such chemical influences. Physically,
alcohols such as methanol are coagulant fixatives, shattering cell structures which formaldehyde preserves intact.
Shattered cells will stain more rapidly than intact ones, and
alcohol fixation may therefore favour formation of the purple complexes. So both chemical and physical factors could
lead to blue, not purple, chromatin following formaldehyde
fixation; such blueness was in fact observed by us (c.f. Wittekind 1979). To decide which factor had the major influence, blood smears were fixed in the following two sequences: formaldehyde followed by methanol, and methanol followed by formaldehyde. The chemical effects of these
two sequences will be similar. However physical effects of
fixatives are influenced mainly by the first fixative to which
the cells are exposed (Horobin 1982, Fig. 2). As both sequences gave blue not purple chromatin, the chemical effect
of formaldehyde was the major influence on staining. Minor
physical influences also occurred however, since completely
shattered, formalin fixed cell nuclei did stain purple.

Applications of the rate control RGE mechanism to other

staining systems
Romanowsky staining has of course been applied to other
materials than smears of human blood and marrow. Several
of these display features readily understood in terms of
the rate control model.
For instance, when G-banding metaphase chromosomes
by Giemsa staining, the chromosomes first appear pale
blue, subsequently displaying the darker purple bands
(Sumner et al. 1973). Rate effects have also been reported
by cytologists using Romanowsky staining. Nuclei in the
poorly accessible centres of a cell clump stain blue, unlike
the purple stained nuclei from the edges of the clump (Boon
and Druijver 1986, p 101). As a final example note that
when smears were prepared from a mixture of blood from
H. sapiens and Amphiuma, Romanowsky staining gave rise
to the usual purple chromatin in the human leucocytes but
only blue staining of the hundred times more massive nonhuman cells (James 1986).
The occurance of the R G E in sections of histological
material can also be rationalised with the aid of the rate
control model described here; however this will discussed
in a later publication.

Romanowsky stains contain the

blue basic dye Azure B plus the
red acid dye Eosin Y

I Uptake of ' f i r s t ' dye by simple I

acid and basic dyeing

It

I BLUE stained basophilic j

sites
I ~

I RED stained acid~philic I

l
sites
il

Uptake of second' dye by

fast staining s i t e s oniy,

| N o short ter,n changes of

|colour
in slow staining s i t e s

yielding Azure B eosinate

precipitates

|
L

~%

f
N

/ PURPLE stained chromatin~

I azurophil ~ d neutrophil ~

/ RED stained eosinopbil \

I granules and erythrocvtes
\I
,

I specific granules, and

II

~\platelets.

/!

/
%~

I BLUE stained lymphocyte

I cytoplasms
x

~l

and nucleoli

/
-

~ g . 1. An outline of the mechanism of

Romanowsky staining

Table 2. Trouble shooting Romanowsky staining *

Problem

Possible causes

A : stain precipitates on slides u

1.
2.
3.
4.

B. overall staining pale,

colour balance satisfactory

1. Dye content too low, due to:

a. Impure batch of dye or stock solution
b. Error in dilution or weighing
c. Precipitation of stain - see A

C: leucocyte nuclei blue, not purple

1.
2.
3.
4.

D. neutrophils appear agranular

1. Any of factors listed in C

E: neutrophils appear " t o x i c "

1. pH too high
2. Staining time too long
3. Azur B concentration too high

F: erythrocytes and eosinophil

granules too blue

1. pH too high
2. Wrong buffer used
3. Staining time too long

Buffer concentration too high

Methanol (or DMSO) content too low
Staining solution too old
Temperature too high

Dye content much too low - see B

Methanol or DMSO content too high
pH too low
Staining time too short

a The ICSH reference method, and normal blood and blood smears, are assumed
b Precipitation can give rise either to: lumps of stain on the slide, see A: 3-4; or to pale staining
resulting from loss o f dye from solution, see A : 1-3

336

Some scientific and technical conclusions

The essentials of our views concerning the mechanism of
R o m a n o w s k y staining are summarised in Fig. 1. These conclusions are used to provide everyday advice, see Table 2,
on problems arising when interpreting and trouble-shooting
R o m a n o w s k y staining.

Acknowledgements. This paper was inspired by the work on Romanowsky staining carried out by Prof. D. Wittekind; who in
addition provided us with samples of pure thiazine dyes. We also
acknowledge Dr. D.J. Goldstein and Prof. A.R. Rogers for the
provision of facilities, and the Berkshire Education Authority for
a subsistance grant to one of us (K.J.W.). The standard azure
B-eosin Y stain was provided by the Community Bureau of Reference of the European Economic Community.
References
Baker JR (1958) Principles of biological microtechnique. Methuen,
London
Bentley SA, Marshall PN, Wade MJ, Galbraith W (1979) Azure
B-eosin staining of blood cells. The effects of variation in stain
formulation and staining technique on stain performance. Anal
Quant Cytol 1 : 179-186
Bessis M (1973) Living blood cells and their ultrastructure. (Translated by RI Weed) Springer, Berlin Heidelberg New York
Boon ME, Druijver JS (1986) Routine cytological staining techniques, theoretical background and practice. Macmillan, London
Clark G, Kasten FH (1983) History of staining, 3rd edn. Williams
and Wilkins, Baltimore
Comings DE (1975) Mechanisms of chromosome banding. IV.
Optical properties of the Giemsa dyes. Chromosoma 50:89 110
Elliott K, Knight J (eds) (1975) Biochemistry and pharmacology
of platelets. Ciba Foundation Symposium, New Series No. 35.
Elsevier, Amsterdam
Fedorko ME, Morse SI (1965) Isolation, characterisation, and distribution of acid mucopolysaccharides in rabbit leucocytes. J
Exp Med 121 : 39 48
Goldstein DJ (1963) An approach to the thermodynamics of histological dyeing, illustrated by experiments with azure A. Qu J
Microsc Sci 104:413-439
Goldstein DJ (1965) Relation of effective thickness and refractive
index to permeability of tissue components in fixed sections.
J R Microsc Soc 84:43-54
Hardin JH, Spicer SS (t 971) Ultrastructural localisation of dialysed
iron-reactive mucosubstances in rabbit heterophils basophils,
and eosinophils. J Cell Biol 48:368 386
Horn RG, Spicer SS (1964) Sulfated mucopolysaccharides and
basic protein in certain granules of rabbit leukocytes. Lab Invest 13 : 1-15

Horobin RW (1974) A preliminary quantitative study of the 'rate

of staining' model for rationalizing various mixed acid dye
stains (' trichromes'). Proc R Microsc Soc 9:110
Horobin RW (1982) Histochemistry: an explanatory outline of
histochemistry and biophysical staining. Gustav Fischer, Stuttgart; Butterworths, London
International Committee for Standardisation in haematology
(1984) ICSH Reference Method for staining of blood and marrow films by azure B and eosin Y (Romanowsky-Giemsa stain)
Br J Haematol 57:707-710
James J (1986) Paper delivered at The First Course in Standardisation and Quantitation of Diagnostic Staining in Cytology,
Breda, The Netherlands
Lillie RD (1943) Some experiments on the Romanowsky staining
of blood films. J Lab Clin Med 28:1872-1875
Lillie RD (1944) Factors influencing the Romanowsky staining
of blood films and the role of methylene violet. J Lab Clin
Med 29:1181-1197
McCall CE, Katayama I, Cotran RS, Finland M (1969) Lysosomal
and ultrastructural changes in human "toxic" neutrophils during bacterial infections. J Exp Med 129:26~282
Marshall PN (1978) Romanowsky-type stains in haematology.
Histochem J I0:1-29
Marshall PN (1980) Romanowsky and reticulocyte stains. In:
Schmidt RM (ed) CRC Handbook Series in Clinical Laboratory Science. Section I: Hematology, vol 2: CRC Press, Cleveland, pp 63-77
Perrin DD, Dempsey B (1974) Buffers for pH and metal ion control. Chapman and Hall, London
Proctor GB, Horobin RW (1986) A widely applicable analytical
system for biological stains: reverse phase thin layer chromatography. Stain Technol 60 : 1-6
Spitznagel JK, Dalldorf FG, Leffell MS, Folds JD, Welch IRH,
Cooney MH, Martin LE (1974) Character of azurophil and
specific granules purified from human polymorphonuclear
leukocytes. Lab Invest 30:774-785
Sumner AT (1980) Dye-binding mechanisms in G-banding of chromosoms. J Microsc l t9:397M06
Sumner AT, Evans HJ, Buckland RA (1973) Mechanisms involved
in the banding of chromosomes with Quinacrine and Giemsa.
II. The interaction of the dyes with the chromosomal components. Exp Cell Res 81:223-236
Wittekind DH (1979) On the nature of Romanowsky dyes and
the Romanowsky-Giemsa effect. Clin Lab Haematol 1 :24~262
Wittekind DH (1983) On the nature of Romanowsky-Giemsa staining and its significance for cytochemistry and histochemistry:
an overall view. Histchem J 15:1029-1047
Wittekind DH, Kretschmer V (1980) Zur Histochemie der Leukozytengranula - Untersuchungen mit reinen Farbstoffen, insbesondere an den neutrophien Granula. Acta Histochem (Suppl)
21 : 39.45
Zanker V (1981) Grundlagen der Farbstoff-Substrat-Beziehungen
in der Histochemie. Acta Histochem (Suppl) 24:151-168