PROEFSCHRIFT TER VERKRIJGING VAN
DEN GRAAD VAN DOCTOR IN DE WIS- EN
NATUURKUNDE AAN DE RIJKS-UNIVERSI-
TEIT TE UTRECHT OP GEZAG VAN DEN
RECTOR MAGNIFICUS J. F. NIERMEYER,
HOOGLEERAAR INT DE FACULTEIT DER
LETTEREN EN WIJSBEGEERTE, VOLGENS
BESLUIT VAN DEN SENAAT DER UNIVER-
SITEIT TEGEN DE BEDENKINGEN VAN DlC
FACULTEIT DER WIS- EN NATUURKUNDE
TE VERDEDIGEN OP MAANDAG 22 JANUARI
192a, DES NAMIDDAGS TE 4 UUR

Militaire verplichtingen hebben mijn studententijd, nu ruim
acht jaren geleden, tot een abrupt einde gebracht, en langen
tijd iederen regelmatigen arbeid belet. Later nam mijn maat-
schappelijke werkkring mij in steeds toenemende mate in beslag,
zoodat het thans voltooide werk soms maanden achtereen moest
blijven rusten. Slechts dan toch achtte ik mi] gerechtigd daaraan
te arbeiden, wanneer aan alle andere aanspraken op mijn
werkkracht ten volle was voldaan. Het late verschijnen van dit
proefschrift moge hiermede gerechtvaardigd zijn.

Dankbaar erken ik de groote waarde, die de academische
vorming voor mij heeft gehad, en gedenk hierbij in de eerste
plaats de hoogleeraren in de Wis- en Natuurkundige faculteit,
wier onderwijs ik aan de Amsterdamsche Universiteit genoten
heb, In het bijzonder U, Hooggeleerde Hügo de Vries, ben ik
veel verschuldigd. Uw leiding, Uw belangstelling in mijn studie,
en vooral ook het voorrecht, dat ik een tijdlang Uw assistent
heb mogen zijn, worden door mij steeds in dankbare herinnering
gehouden.

Hooggeleerde Went, het verheugt mij hier een gelegenheid
te hebben openlijk te kunnen gedenken de bereidwilligheid,
waarmede gij op het verzoek om als mijn promotor op te
treden, zljt ingegaan; de liulpvaardigheid, waarmede gij mij
liet materiaal voor mijne onderzoekingen hebt verschaft; en de
welwillendheid, waarmede gij mij bij de bewerking van dit
proefschrift in alles zijt tegemoet gekomen. Voor Uw moeite.
Uw tijd. Uw voorlicliting en Uw belangstelling l)en ik U
blijvend erkentelijk.

\'quot;liuioc«»^nbsp;\'tiir-\'innb\':. !.)«!«/, \'«^f » nji^-nbsp;\'■ TMrf

iS\':.\'.nbsp;■nbsp;»■•.■iM (j\'fv •il\'-J.-iii\'«».\' .V^ämW «fr. \' Ti . \'\'\'\'

. M-i« -n.^ \'^i-nAfjti^ü-f Jii\'.\'.finbsp;.\'.-quot;l/yquot; \' ïr ••jnKif« 4»t;f .

0
7
18
37
50

59
01

List of literature.............

IV.nbsp;Systematical survey of atypical embryosacs . .
V.nbsp;ConclusionH regarding systematics and phylogony

embryosac development
VII. Abnormal sacs which should fail to follow tho out
lined system
, VIII. Summary



M
• - \' .
i.

quot;jîz •■ .. ..nbsp;i.0» irtiiisH^v»»?:-. \'.».î.rfîi - irüfïótv^\') .V-

Moringa oleifera Lam. (pterygosperma Gilrtner) is the best
known representative of the tropical family of the Moringaceae.
Indigenous to the Indies the species has been cultivated for
many thousand years and is to l)e found in African and American
tropics as well as in Asia.

One of the very last studies published by Tiieub (Le sac
embryonnaire et l\'embryon dans les Angiospermes. Nouvelle
série de recherches —Ann. du Jard. Bot. de Bnitenzorg 1910) was
meant as the first of a series of articles on the subject. For
this purpose he bronglit liome an extensive innterial relating
to embryosac formation in numerous tropical families. After-
wards the collection passed to Prof. Wknt of Utrecht University
who kindly put at my disposal the Moringa material collected
l)y TiiF.un at Buitenzorg (Java), which was fixed in alcohol.

From still two other sources material was available at
Utrecht. Boldingii collected on the isle of Curacao (Dutch West
Indian archipelago) and Kuypeii in Surinam (Dutch Guyana).
Fixation by both in alcohol and in Flemming\'s.

During the examination it turned out that only embn/osac
formation could be studied from tho material at hand. Even in
seeds of considerable size no trace of embryo formation was
seen, though full-grown seeds were known to carry quite nor-
mall embryo\'s. Dr. Stauki. at Paramaribo (Surinam) kindly
helped me by sending additional stadia, collected by him and
fixed in alcohol, which fixation proved itself to be the best.

Though of different origin the material was of absolute uni-
formity as to megaspore- and embryosac formation. Kuyper\'s
material however showed some delay in development. Ovaries
of a size in which usually mature sacs are found, contained
but tetrads or exceptionally a two-nucleate stage. This delay
is distinct from the beginning, the nucellus remaining without
any differentiation until an exceedingly advanced (Plate I,
Fig. 3) and far later stage than usual in all other material.

The examination of many thousands of sacs underlies this
publication, but never anything has been seen, which was not
in conformity with the following description of the development
of the female gametophyte.

Ordinarily the archesporium is distinguishable about the time
of the first differentiation of the integuments. It is onecelled
and hardly to be recognised from the cells of the surrounding
nucellar tissue. (Plate I, Fig. 1). Occassionally a two celled ar-
chesporium is met with (Plate I, Fig. 2), both cells showing the
same germinating capacities, which might lead to two tetrads
and even to two complete embryosacs lying paralel.

The archespore cell does not divide and is the embryosac
mothercell (Plate I, Fig. 4). Parietal tissue is totally suppressed,
but exceptionally archespore cells containing two nuclei are
met with. (Plate I, Fig. 5, G). In no case were cell walls seen.

The embryosac mothercell gives rise to two cells of unequal
size, a large inner one and a small outer cell (Plate I, Fig 7). Tn
the second division the spindle of the outer cell lies rectangular
to the axis of the sporangium. Thus the four megaspores are
never found in a row, but always the two outer cells at a
right angle with the inner ones (Plate I, Fig. 8).

Megaspore formation is followed by a rapid désintégration and
final disappearance of the outer cells. The first signs of destruc-
tion are already seen when the functioning inner one is still one-
nucleate and has hardly begun to grow (Plate I, Fig. 9) and even
before the end of the two-nucleate stage has been reached their
disappearance is complete (Plate T, Fig. 12, Plate II, Fig. 13).

The inner megaspore which from the very first moment shows
itself the functioning one, is regularly filled up with cytoplasm.
Its nucleus is to be tound at the top end, where it divides
(Plate I, Fig. 10, 11). Soon after this first division one of the
daughter nuclei commences to move to the lower end of the
sac. This migration is accompanied by an ever increasing polari-
sation, the result being the typical and well know figure of the
polisared two-nucleate stage of the embryosac: two nuclei separa-
ted by a large central vacuole (Plate I, Fig. 12, Plate II, Fig. 13,14),
and both embedded in a comparatively small mass of cytoplasm.

By a second division the four-nucleate stage is reached.
(Plate II, Fig. 18, 19). The primary micropylar nucleus however
seems to be in advance (Plate II, Fig. 15), sometimes even as
much as having finished its division when the primary chalazal
nucleus is still at rest (Plate II, Fig. IG, 17). Embryosacs are
then three-nucleate.

The four-nucleate stage is followed by a division of one of
the nuclei only, both chalazal and one micropylar nucleus
remaining undivided. So at the top end never more than three
and at the chalazal end never more than two nuclei are seen.
(Plate III, Fig. 20, 21, 22). The egg apparatus is formed in the
usual way, the three nuclei get separated by cell walls (Plate III,
Fig. 23) and finally the two well shaped synergids partly cover
the egg (Plate 111, Fig. 24). In the mature sac cytoplasm in the
cells^ of the egg apparatus shows the normal distribution. As
usual the synergids are characterized by a large vacuole at
the lower end, cytoplasm and nuclei being gathered at the topcnd.

During the formation of the egg apparatus the two chalazal
nuclei are seen in close connection and steadily moving upwards
(Plate 111, Fig. 21, 22, 23). Finally a position is reached at the
very topend of the embryosac quite close to egg and synergids
(Plate 111, Fig. 24). There has never been seen any sign of fur-
ther division or of fusion.

Thus the mature embryosac is five-nucleate, three ot the
nuclei being of micropylar and two of chalazal origin. The

synergids are sisters, and an upper polar nucleus is lacking.
Two chalazal nuclei have taken the position of an embryosac
nucleus, and antipodals are missing.

Fertilization takes place in a normal way, the pollentube
discharging its contents in one of the synergids (Plate III, Fig.
25). One male nucleus fuses with the egg nucleus and the second
male nucleus moves towards the «embryosac nucleus« (Plate III,
Fig. 26). Up to this very moment the latter has retained its
double character and so three nuclei are seen fusing, giving
rise to the primary endosperm nucleus.

Endosperm formation commences at once and soon a great
many nuclei are present. They are especially numerous at the
top end and some are found all along the walls of the em-
bryosac. All the time the sac is rapidly increasing in size, but
the fertilized egg shows no signs of any activity. Not until
the seed has reached a length of about 4 ram. does the egg\'s
first division take place (Plate VI, Fig. 35: mature sac — and
sac after second division of egg).

Embryo development begins with free nuclear division. The
egg nucleus divides without cell wall formation giving rise to
a two-nucleate embryo (Plate IV, Fig. 27). This first division is
followed by a simultaneous division of both nuclei, their
spindles at a right angle. The four nuclei resulting from this
division are not lying on the same level. Somitimes they are
found in two succesive sections (Plate IV, Fig. 28) and when in
one figure (Plate V, Fig. 29) they arc still distinctly on separate
levels. The free nuclear division goes on till a sixteen-nucleato
embryo stage is reached (Plate V, Fig. 30). Then walls are formed,
the 16-nucleate embryo thus developing into a IG-celled one
(Plate V, Fig. 31). Simultaneous divisions have come to an end
now, seven of the sixteen cells figured, containing one nucleus
each, and eleven of them showing two nuclei.

The endosperm is still without cell walls, the free nuclei
lying embedded in a common mass of cytoplasm (Plate VI, Fig.
33, 34). Cytoplasm however becomes more and more vacuo-

lated and shortly afterwards is found divided by cell walls.
(Plate VI, Fig. B2). At this stage seeds have already attained
a length of about 8 mm. and the fruits even of 15 to 25 cM.

Four megaspores are formed, the outer two lying rectangular
to the axis of the sporangium.

The inner megaspore is the functioning one. By two suc-
cesive divisions the normal polarised four-nucleate stage is rea-
ched. The third division is restricted to one of the micropylar
nuclei only, the other micropylar and both chalazal nuclei

Thus the mature embryosac is five-nucleate, showing a nor-
mal egg apparatus of two synergids (sisters) and the egg. The
position of the embryosac nucleus is taken by the two chalazal
nuclei, lying iu close contact and quite near the egg.

At fertilization one male nucleus fuses with the eggnucleus
and the other male nucleus enters in triple fusion with the
chalazal nuclei. This primary endosperm nucleus at once com-
mences to divide and soon numerous free endosperm nuclei are
present. The egg remains undivided for quite a long time.

Embryo formation begins with free nuclear division up to the
sixteen-nucleate stage. Then cell walls are formed and further
divisions no longer take place simultaneously. The embryo
rapidly grows and the cytoplasm in which the endosperm luiclei
are\'embedded is divided by cell walls.

This way of megaspore — and of embryosac formation, lea-
ding to a u-nucleate sac of a very special character, up till
now has only been reported for Oarcinia (M. Trkoh: Ann. du
Jard. Bot. de Buitenzorg 1910: Le sac embryonnaire et I\'eni-
bryon dans les Angiospermes. Nouvelle s6rie de recherches, I
Garcinia Kydia (Roxu.), Garcinia Treubii (Pieurk)).

Up to the beginning of this century hardly any attention
was paid to the study of the development of the Angiosperm
embryosac. Only veiy few cases of\'\'abnormal sacsquot; being known,
it was generally accepted that the development was of a most
striking uniformity throughout the whole group. This normal
course runs as follow: By two successive divisions the embryosac-
mothercell gives rise to a row of four cells, called megaspores.
Three of these soon begin to degenerate while the fourth, ra-
pidly increasing in size, becomes the functioning embryosac.
Its nucleus divides itself thrice thus producing eight nuclei,
originally free in the same plasm but soon separated by cell-
walls. These eight nuclei are arranged into two groups: viz.
a micropylar one (egg and two synergids) and a chalazal one
(three antipodals). In the middle of the sac the two remaining
nuclei (polar nuclei) are seen fusing (embryosac-nucleus).

The last fifteen years however have brought to light an ever
growing number of atypically developing sacs. Deviations in
almost every direction were detected. First of all the number
of megaspores seems to vary fiom the normal four to only one
(quot;row of threequot;, quot;row of twoquot; or quot;embryosac-mothercell func-
tioning as embryosacquot;). Secondly the number of nuclei in the
fullgrown sac is far from being regular. Instead of the usual
eight there may be sixteen or only four; not to speak of the
numerous cases in which a secondary increase or decrease of
nuclei could be stated.

cies or families, while others affect representatives of widely
separated groups. Very soon the question rose whether these
irregularities ought to be considered as more primitive or as
more advanced than the normal type. The origin of the Angio-
sperm embryosac is still utterly unknowii and it was hoped
that some light might be given by the study of «abnormal
sacs«. So gradually the attention became focussed on the pro-
blem: how to arrange the deviations from the normal type in

If only a survey of the material was wanted we should be
free to choose one or two of the most prominent characteristics
as a base on which a scheme could be built. Since however an
insight in phylogenetical problems is wanted, we are bound to
reckon with all the facts that cooperate in embryosac-formation.
There can be no objection to stating that artificial systeuis are
entirely worthless for phylogenetical purposes. A useful system
of the gametophyte necessarily must be natural.

We will try to give a complete survey of all atypical em-
bryosacs and of the various systems in which they are classed
by difterent authors. At the same time we will have full oppor-
tunity of discussing any questions of homologizing of the nuclei

It is not the intention to give anything but the outlines of
a natural system. Our knowledge of euibryosac-foruiation in
most families is still too defective to enter into particulars.
Therefore those facts of which the systematical value is clear and
which do not present any points for divergence of oi)inion, are
only mentioned without further dicussion. So for instance whether
the arrangement of the megaspore cells is j or V or whether
the inner one or any of the other three becomes the functioning
megaspore, whether one or more of these megaspore cells develop,
etc^ are details of systematical interest, for which reference to
special studies on these subjects will answer.

It would bring us too far to discuss all literature on the
origin of the Angiosperm embryosac. We will confine therefore

this critical review to those publications which consider also
the systematical side of the question.

As stated above there is often, instead of the row of four
megaspores, a row of three, a row of two, or even no row at
all. A vivid discussion was started in 1908 by Ernst (1908a,
1908(5) on the one and Coulter (1908) on the other side on the
question as to which nuclei in these cases ought to be regarded
as the homologues of the megaspores. Two hypotheses were
suggested and held up to the present moment. Fig. \\a and 16
illustrate the two opposite opinions, clearly showing the diffe-
rence between Coulter\'s and Ernst\'s view.

As most authors refer to these two conceptions we will dis-
cuss them more fully in essence and consequences.

1°. According to Coulter the nuclei produced by the second divi-
sion of the embryosac-motkercell-nucleus are always lo be regar-
ded as nuclei of megaspores. Usually these megaspore-nuclei
are separated by cellwalls, the developing embryosac thus
being of quot;monosporicalquot; origin. Occasionally however the second
division, or even the first division too, is not accompanied
by cellwall formation. In these cases both or all four mega-
spore-nuclei may develop, the results being quot;bisporicalquot; or
quot;tetrasporicalquot; embryosacs.

Accepting the consequences of this point of view. Coulter
further defends the following principles:

2°. Megaspore-formation w determined by chromatine-reduction. T\'he
megaspore nucleus is the first nucleus of the gametophyte,
i. e. of the n-generation. The end of the 2n-generation and
the beginning of the n-generation is determined by the pro-
cess of chromatine reduction. As long as chromatine reduc-
tion did not occur the nuclei still belong to the sporophyte
and they can not be called megaspores.

3°. Megaspore-formation is entirely independent of cellwall-forma-
tion. Cellwall-formation is of only small importance for homo-

logising purposes. Unquestionable cases of desintegratiug cell-
walls are known as well as cases of total suppression of the
megaspore-cellwalls. For instance Mc Allister (1909, 1914)
states that cellwalls originally formed between the mega-
spores of Smilacina and of some of the other Convallariaceae,
soon break down and finally disappear. And Smith\'s (1911)
description of Clintonia shows four raegaspores in the same
cell, there being never more than a small trace of cellplate-
formation.

4®. The number of nuclei in the fulkjvown sac his no phylogenetical
or systematical significance at all.

Fig. Irt shows plainly enonght that the 8-nucleate stage of
a quot;tetrasporical sacquot;, the i-nncleat« stage of a quot;bisporical sacquot;,
and the 2-uucleate stage of an ordinary quot;monosporical sacquot;
are considered to represent the same stage of development.
So there is no sense in phylogenetical hypotheses or in
systems, based on the number of nuclei.
5°. The number of divisions from the embrijosac-mothercell up to
the egg furnishes the most valuable data for phylogenetical and
systernatical studies.

One of the most striking facts in the evolution of the
vegetable kingdom is the continuous shortening of the ga-
motophytc-generation. Among Gymnosperms numerous divi-
sions of the megaspore-nucleus are still prevailing. Within
the Angiosperm group however this number has been reduced
greatly. Normally it is only three, making a total number
of five divisions from the embryosac-mothercell up to the
egg. A total of four or three divisions is also known and
theoretically the possibility of only two divisions (megaspore-
nucleus — egg: the animal condition) must be admitted.
The smaller the number the more advanced the type. The
occurence of more than five divisions on the other hand
should indicate a more primitive condition.

It must be observed here that neither the idea of identifying
megasporn-tormation with chromatine-reduction, nor the idea
of using the number of divisions for phylogenetical purposes,

were new at the time of Coulter\'s publication. Miss Pace (1907)
in her study on Cyprepedium pointed out how the embryosac
with its three divisions between mothercell and egg, was well
on the way to the animal condition. And the coupling of
megaspore-formation to chromatine-reduction is yet met with
in Schniewind-Thies\' (1901), and propagated by Davis (1905),
Chamberlain (1905), and Pace (1907). In fact all the ideas more
fully developed by Coulter were already underlying Miss Pace\'s
publication. She made exactly the same homologies and even
saw the necessity of discerning mono-, bi- and tetrasporical sacs.

After Coulter\'s lucid statement most authors who dealt with
the subject accepted his views and propagated his opinion, among
them Stephens (1909«, 19096), Mo Allister (1909, 1914), Pace
(1909), Smith (1911), Brown and Sharp (1911), Sharp (1912),
Dahlgren (1915), Kusano (1915), Palm (1915), Häuser (191G)
and IsHiKAWA (1918).

Brown (1908, 1909) however did not wholly agree with
Coulter. He admits neither chromatine-reduction nor cellwall-
formation as a criterion for megaspore-formation. According to
him the appearance of cell-plates in the spindle figures of the
first divisions furnishes the only certain characteristic of mega-
spore-formation. As long as cell-plates are present we have to
do with spore-formation; when they are lacking, spore-germi-
nation is going on. This conception however can be regarded
as wholly miscarried since from several sides attention was
called to the fact that cell-plate-formation may occur at any
stage of the embryosac-development, and apparently without
any connection to megaspore-formation.

Dahlgren\'s (1915) attempt to outline a scheme which should
embrace all cases of atypical embryosacs, is still rather primi-
tive. He discerns four groups. The first one, showing five divi-
sions between embryosac-mothercell and egg, is represented by
the normal eight-nucleate sac. Next comes a group with only
four divisions, including the IG-nucleate Penaeaceae, the 4-nu-
cleate Onagraceae, as well as Clintonia, Codiaeum and Lawia.

Only three divisions are found in Podostemon, Dicraea, Cypri-
pedium, Helosis and Statice. While the last group (two divi-
sions == the animal condition) shows a reduction to the utmost,
as seen in Plumbagella.

It will be noticed that this quot;sj^stemquot; is thoroughly one-sided.
All possible stress is laid on the number of divisions, without
giving any thought at all to the number, the arrangement, or
the origin of the nuclei in the mature sac. Not even the
origin of the sacs themselves is regarded, for representatives of
Coulter\'s mono-, bi-, and tetrasporieal sacs are readily joined
in the same group. Of the naturalness and phylogenetical signi-
ficance of a system like this nothing needs to be said.

Samuels\' (1912) scheme is a little more advanced. He too
accentuates the number of divisions and uses them as main line
for his system. (To Dahlgren\'s four groups he added a fifth,
with six divisions and thus more primitive than the normal
type. This new group liow^ever can be dropped safely, since
DessiatofI\'-\'s (1911) observation of a quot;monosporicalquot; IG-nucleate
Euphorbia proved to be wrong). As a new factor Samuels in-
troduced a subdivision ot the groups by means of the mono-,
bi-, or tetrasporieal character of the sacs. In his system not only
the number of divisions, but also the origin of the embryosac
is reckoned with. Too many points however are still left out of
consideration. Though improved the system remains artificial.

Palm (1915) who in a more extensive study advocates the
same scheme rightly remarks: quot;Das diese Aufteilung nur eine
künstliche sein kann ist ja selbstverstilndlich.quot; Its lack of phy-
logenetical value is best shown by reproducing Palm\'s scheme,
in which the dillerent types are called after their first repre-
sentative known:

IsHiKAWA (1918) lately published a scheme which from a
phylogenetical point of view is certainly to be marked as a
distinct progress. He no longer sticks to one or two rather
voluntarily chosen moments in the development of the embryo-
sac, but he tries to reckon with the other facts as well. His
scheme (Ann. of Bot. 32, p. 305, Fig. XI) not only deals with
the number of divisions and with the number of megaspores
which join in embryosac-formation, but also pays attention to
the origin and to the number of nuclei in the full-grown sac.
So it really contains some necessary elements for the building
up of a natural system.

When put into practice however its usefulness is rather
limited. It meets our present knowledge of embryosacs, but is
not planned broadlj\'^ enough to include further possibilities. It
gives an insight in the author\'s views on homologies, but it
does not give a valuable system. It is meant to give much,
but it is worked out confusedly. Perhaps that is the reason why
all harmony with the sporophytic system is absolutely lacking.

As the most succesful arguments against the view, the
development which of we just finished sketching, the following
has been brought forward:

1°. Against the assumption that megaspore-formation cannot be
shortened and always must be preceded by two divisions,
was moved the fact that the sporogenous tissue has gra-
dually been restricted from an elaborate tissue among Gym-
nosperms to only one cell in most Angiosperms. There is
no reason why this tendency to shorten the gametopliyte-
generation should have stopped there. On the contrary we
might expect this tendency to go on and affect megaspore-
formation. It gives a natural explanation of the reduction
series as demonstrated in the quot;row of fourquot;, quot;of threequot;,
quot;of twoquot;, quot;no row at all.quot;
2°. Against the assumption that megaspore-formation should be
determined by chromatine-reduction, was moved Muhbkck\'s
(1901) discovery of the embryosac-development in Alche-

milla, in -svliicli species a row of four is quite normally
formed, without any reduction in the number of chromo-
somes. It seems hardly possible not to homologise this row
of four with the ordinary row of four megaspores, especially
since one of the four develops to a normal 8-nucleate embryosac.
\'6°. Against the assumption that cellwall-omission should induce
the development of two or more megaspores in the same
cell, was moved the fact that no cases are known of four
unquestionable megaspores, lying in the same cell and de-
veloping all four to form one embryosac of the usual 8-nu-
cleate type. On the contrary it is difficult to see why four
developing megaspores should arrange their nuclei into tivo
groups (a micropylar and a chalazal one), just as in an or-
dinary monosporical sac and without leaving any trace ot
the tetrasporical origin.
4®. Against the assumption that the number of nuclei in the
full-grown sac should be of no importance at all, was moved
the fact that the 8-nucleate sac is of such remarkable fre-
quency, that its appearance cannot be believed to be mere
chance. The less so since (according to Coui-teii) these eight
nuclei represent either the greatgranddaughters (in mono-
spor. sacs), or the granddaughters (in bispor. sacs), or the
daughters (in tetraspor. sacs) of the megaspore-nuclei.

Eunst (1908a, 190Sb) rejecting Coulter\'s view and all clas-
sification based on the number of divisions points out that:
1°. Two distinct processes can be recognised in the life history
of the gametophyte, viz. enün-yosac-formation and embryosac-
(levelopmcnt. quot;Die Entwicklungsvorg{lnge im Embryosack
scheinen mir unabhängig von seiner Entstehung betrachtet
werden zu nülssen.quot; quot;Die fünf Teilungen gehören ja ganz
verschiedenen Entwicklungsvorgilngen an.quot; \'\'Die beiden ersten
repräsentieren die letzten Teilungen in einem Makrosjwrangium...
und gehören dem Vorgang der S ore wbildung an.quot; quot;Die
drei anderen Teilungen dagegen erfolgen im Verlaufe der
Sporentoi?m^quot; (1908b, S. 2G).

reduction. Instead of four, only three or two megaspores
are formed, or even the embryosac-mothercell itself func-
tions as megaspore. Thus chromatine-reduction, ordinarily
occurring during embryosac-formation, necessarily is trans-
ferred to a later stage. quot;Bei teilweiser Unterdrückung der
Tetradenteilung wird der zweite Teilungsschritt der Reduk-
tionsteilung in die keimende Spore verlegt, und bei voll-
ständig ausbleibender Tetradenteilung finden beide der zur
Beduhtion notwendigen Teilungen innerhalb der keimenden Ma-
hrospore statt.\'\' (S. 27).
3°. The process of embryosac-Jßw/öpme??^ is wholly independent
of that of embryosac-formation. Among Liliaceae e.g. all
types of megaspore-formation are found, from the normal
tetrad down to total suppression. Always however the func-
tioning megaspore — by three divisions — reaches the 8-
nucleate stage.

The process of Qm\\gt;xyo^2iC,-development is determined by the
number of divisions, by the arrangement of the nuclei, by
vacuolation and by cell-formation.

On these grounds Eunst, when reviewing the literature, con-
cludes that two types of embryosacs can be recognised, viz. the
ordinary eight-nucleate one and a sixteen-nucleate type, quot;als
altere oder doch als selbstständige Form des Embryosackes der
Angiospermen« (S. 29). This more primitive type is distinguished
by the divisions numbering four instead of three, by the ab-
sence (at least at first) of a central vacuole, and by the lack
of bipolarity. The 16-nucleate sac thus shows itself to be of a
primitive character as to embryosac-development, and of a
reduced nature as to embryosac-formation, there being no row
at all, the embryosac-mothercell itself functioning as embryosac.

This marked distinction between embryosac-formation and
development is a real advantage on Coultkr\'s system. It is a
first step on the way of treating the various processes of the
female gametophyte separately. So far it opens the prospect
of getting a natural system. Laying all the stress however on

the total number of nuclei in the mature sac, without
paying any attention to their origin, makes Ernst\'s system
almost as artificial as Coulter\'s. For there are still many more
factors, which show an independent line of development in the
life history of the gametophyte. That is why a lot of abnormal
embryosacs, discovered since Ernst published his system, could
not be placed in his scheme.

To complete this review of systematical and phylogenetical
studies on Angiosperm embryosacs, we have still got to men-
tion the publications of Campbell, of Jacobsson-Stiasnv and of

Campbell (1899, 1900, 1902, 1903, 1905, 1909, 1910, 1911, 1912)
in a series of articles tried to propagate the idea that quot;the
embryosacs with an increased number of nuclei are older ty-
pesquot; (1911). It is only the numhcr of nuclei that counts with
him; their ovujin is not thought worth much attention. Accor-
ding to. him there is a gradual passing on from the multicel-
lular Oymnosperm type to tho ordinary S-iuicleate Angiosperui
sac: Pandanus with its 32—64 antipodal nuclei, is quot;really pri-
mitivequot;, and the 16-nucleate sacs of Pei)eromia and Cuinnera
form the transition to the normal type.

Emma Jacobsson Stiasny (1916) rightly states quot;im Cegensatz
\'AW Ernst, dasz die Anzahl der Kerne des reifen Embryosackes,
im Gegensatz zu Coultkk, dasz die Anzahl der Teilungen allein
noch nicht zur Charakterisierung der Stellung genügen kann.quot;

quot;kausalmechanische Darstellungquot; however is at least as
one-sided ns any of the older systems. The whole study is
based on the idea of quot;Ernahrungsverhaltnissequot; being the only
possiI)le cause ol any atypical number of nuclei in the embryo-
sac. Far too much importance ia attributed to the number of
nuclei in the i\'ull-grown sac, and scarcely any attention is paid
to their origin. Without further investigation tho 16-nucleate
sacs are accepted to represent a type of their own.

SciiOuHOFr (1919) is only mentioned here because his publi-
cation is of so recent a date. It will do to state:
1°. That his system is hmed on the absolutely false assumption

that one synergid should be a sister to the egg and the
other to the upper polar nucleus! All students however
agree on both synergids being sisters, but the author, ap-
pearing to throw aside his usual powers of self-criticism,
moves as conclusive proof in favour of his view: quot;Da nach
meiner Erklärung die eine Synergide eine Schwesterzelle

2°. That lots of his further arguments are taken from publica-
tions over fifteen years old and which have never since been
confirmed.

3°. That his arguments are sometimes misleading as he cites
from preliminary notes which have been rectified later on
(e.g. his quotations from Campbell on Pandanus, 1909, and
from Stephens on Penaeaceae, 1908, revised by these authors
resp. in 1911 and in 1909).

Our review of literature has led us to discern two sets of
systems, the one based on the numl)er of divisions from em-
bryosac-mothercell up to the egg, and the other on the number
of nuclei in the full-grown sac. All of these schemes however
were artificial and none of them succeeded in giving an insight
in origin and phylogeny of the Angiosperm sac.

Before beginning our attempt at a natural system of the
gametophyte it will be good to remember sporophytical condi-
tions which show that each part of the plant follows its own
line of development. That is why descent never can be stated
with absolute certainty, phylogeny always depending on a com-
pJex of data. This, of course, must be applied to the study of
the gametophyte as well. A system based on the number of
nuclei in the full-grown sac (Ernst) or on the number of divi-
sions from mothercell to egg (Coulter) can hardly be considered
of greater value for phylogenetical purpose.? than a sporophy-
tical system based on the number of anthers only. If we want
to detect the relations between the eight-nucleate sac and the
abnormal ones, we must make clear first how many individual
morphological characters can be recognised in the female game-

tophyte. Usually this gametophyte is considered a morphological
unit. One can hardly deny, however, that it is of a complex
nature and that its morphology has to reckon with the following
processes :

If we want to obtain results of phylogenetical value we have
got to study the morphology of each of these processes in detail.
Coulter and his school identified chromatine-reduction and mega-
spore-formation, and likewise Ernst mixed up two lines of deve-
lopment when basing his system on the total number of nuclei,
without paying any attention to their origin. Of course it is
quite possible that there exist connections like those suggested
by Coulter and Ernst, but there is no good in presupposing
them. If they exist they will come to light even when treating
the processes separately.

We will first discuss chromatine-reduction and polarisation.
These two seem to be processes of great constancy and are very
seldom, if ever, affected by deviation from the normal.

For years this process has been one of the main objects of
cytological research. It is not necessary to give a description
in detail of the phenomenon, the more since it is apparently
without any phylogenetical value. With the few exceptions,
presented l)y apogamous plants, reduction division always occurs
immediately after the formation of the embryosac-mothercell.
The process is not adected l)y deviations in megaspore forma-
tion, and seems to bo of the utmost constancy. It goes on in
eml)ryosac-mothercells, resulting in four megaspores, as well as
in embryosacs derived from the mothercell without any mega-
spore formation.

polarisation is, lias never been subjected to a special study.
Its origin, its inducing factors, its function, etc. are entirely
unknown, which is the more remarkable since everyone who
has ever studied the development of an embryosac must be
familiar with the process. It goes beyond the scheme of this
paper to expatiate on questions connected with its meaning.
From our point of view it will do to state that the process
goes on in all embryosacs in absolutely the same way. The
formation of the well known large central vacuole is the most
prominent phenomenon that accompanies polarisation and there-
fore worth our special attention. In the normal eight-nucleate
sac, sprung from a quot;row of fourquot; no trace of vacuolation \') is to
be seen during megaspore-formation. Protoplasm remains a ho-
mogenous mass until the development of the embryosac (mega-
spore) begins. Then vacuolation commences as a group of small
vacuoles which soon results in the usual large central vacu-
ole. This process passes with such rapidity that even the
two-nucleate stage could never be found without the typical
large vacuole, which separates both nuclei, indicating them
as primary micropylar and primary chalazal nucleus. This
normal course of vacuolation is so common that most authors
do not even mention it. As a matter of fact polarisation
and vacuolation of the embryosac are mostly left out of
discussion, only few authors indicating them. Sometimes even
their figures are but outlined, all information on the subject
thus lacking. Modilewski (1909) seems to liave felt its signifi-
cance when saying: quot;Man musz aber womöglich nicht mu*
dieses Merkmal (die Zahl der Kerne) sondein auch die Höhe
der Symmetrie und die Polarität des Embryosacks in Betracht
ziehen. Dan wird man vielleicht imstande sein, einige Anhalts-
punkte zu gewinnen.quot; No attention at all, however, is paid to
these words. A thorough study of the publications (including
their figures!) on abnormal embryosacs will show the correct-
ness of MoDii.KwsKi\'s remark and the fundamental significance

•1) It is not tlio intention to give any opinion on tlic origin and formation of
tlifi vacuoles. Hero and in tlie followin^r the word quot;vacnolationquot; is only nsod for
indicating the ajjpearence of the central vacuole.

of the process of vaciiolation for hoiiiologising the dilferent
stadia in the development of atypical sacs. The process just
described and well-known to all students of morphology, gives
cause to call the attention to the following points:
1°. Polarisation (vaciiolation) is a function of the emhnjosac (dc-
veloping megaspore). It^ does not accompany the megaspore-
formation, but its development. It commences as soon as megaspore
development begins.

Tliis remarkably constant character, viz. vacuolation just pre-
ceding the first division of the functioning megaspore, furnishes
us with a new characteristic by which megaspores may be
recognised, even when the quot;row of fourquot; is not formed. As long
as plasm remains homogenous, megaspore formation is still
going on. As soon as vacuolation commences, megfispore deve-
lopment has begun. So the nuclei just preceding vacuolation are
to be considered as megaspore nuclei.

I am well aware that this proposed use of vacuolation as
means of recognising megaspores is nothing more than a working
hypothesis. Its progress in the normal sac is no reason in itselt
to assume that vacuolation always and under all circumstances
should be bound to the early stages of spore-germination. There-
fore I will move some arguments in favoui- of the hypothesis.

Firstly in literature no case is met with in which vacuolation
did not commence just after megaspore-formation.

Secondly the hypothesis is confirmed by all well-established
and undoubtable cases of megaspore-fornnition under abnormal
conditions (which fully justifies the application of the idea to
those cases in which homologising meets with difliculties). For
instance: Smith (1911) describes the embryosac-mothercelUnucleus
of Clintonia giving rise to a row of four nuclei, not sei)arated
by cell walls. Three of these soon desintegrate, only the upper
one developing. Nobody will dispute the megaspore-cluiracter of
these four nuclei. Though all four megaspores are lying in the
same cell, phwn remains homogenous up to the first division of
the developing nucleus. According to Joiin.son (1911) the walls
between the megaspores in Peperomia hispidula are very deli-
cate and soon disappear, leaving four nuclei in a continuous

mass of cytoplasm. Dm-ing the preparation for the next division
the large central vacuole is rapidly formed. In Peperoraia Sin-
tensii (Brown, 1908) an evanescent wall appears in the first divi-
sion only; in other Peperomia\'s (Campbell, 1899; Johnson, 1900)
cell wall formation is wholly omitted, but in all these cases
cytoplasm remains homogenous up to the third division (= first
division of the megaspores).

no walls; vacuolation
following third divis.;
tetrasp. embryosac
(e. g. Peperomia).

no walls; vacuolation
following sccond di-
vir.; bisp. embryosac
(e. g. Gunnera)

Thirdly we get results from the application of the hypothesis:
Up till now vacuolation was entirely left out of consideration
by all authors, and embryosacs were considered of identical
development, whether showing vacuolation after the first or
second or third division. So for instance any reasonable expla-
nation of conditions in Gunnera with its seven fusing nuclei and
in Peperomia with its eight fusing nuclei, was absolutely lacking.

When marking vacuolation, differences will he noticed in
the early stages of embryosac-formation in e. g. Lilium, Glun-
nera and Peperomia, and they will be recognised as being of
resp. mono-, bi- and tetrasporical character (Fig. 2). Likewise
the peculiar number of eight fusing nuclei in Peperomia (Fig. 9)
and of seven in Gunnera (Fig. 11) loses its mystery. It is not
necessary to supply further examples here; the hypothesis\'
working capacities will become more and more clear in the
following pages.

2°. bi the tioo-nucleate stage of a normal embryosac the nuclei
are always separated by the large central vacuole (the embryosac is
polarised).

The significance of this central vacuole for homologising pur-
poses is plain. It provides us with means to distinguish the
nuclei of the micropylar end from those of the chalazal end.
There is no difficulty whatsoever in distinguishing the pri-
mary micropylar nucleus and the primary chalazal one. Only
few authors however have realized the significance of this pola-
risation. As we will point to it often later on, one instance
will do for the moment to illustrate its extraordinary value:
In Onagraceae all four nuclei of the mature sac are found
at the micropylar end. The likewise four-nucleate Plumbago sac
shows two nuclei at each end of the central vacuole. In the
first case all nuclei are of micropylar origin, in the second one
two are micropylar and two chalazal. Though both sacs are
four-nucleate it is undoubtedly a mistake to liomologise these
two. From the very beginning of their development they are
plainly different.

Normally a quot;row of four« is formed, one of which becomes
the functioning megaspore, Whether the arrangement of these
cells is : or V or .:•, and whether the inner one or any of the
other three becomes the functioning megaspore \'), »^ud even

whether one or more of these megaspore-cells develop are
questions most probably of systematical interest as well, but
lying beyond the scheme of this study. It is the number of
megaspores formed, that interests us now, and whether it is
possible that two or more raegaspores enter in embryosac-for-
mation, thus affecting the number of nuclei in the mature
sac. When formulating these two points more exactly, it turns
out that we have got to study the following lines of deviation
from the normal type:

It is known and needs no further commentary that the usual
number of megaspores is four. Normally three of these desin-
tegrate as soon as embryosac-development begins. Theoretically
however we might as well expect a partial or total suppression
of these non-functioning nuclei. The development of a reduc-
tion in this direction is fully worked out and represented in
fig. 3. The (normal) «row of four« is figured by A. B shows
the «row of three«, C the «row of two« and in D «the
embryosac-mothercell itself is seen functioning as an embryosac.«
In agreement with this gradual suppression, respectively three,
two, one or none megaspores are seen desintegrating.

It must be emphasized Iiere again that this reduction series
rejects all presupposed connections between megaspore-forma-
tion and chromatine-reduction. Wiien there is a row of four
or a row of three (fig. 3 A, B) chromatine-reduction is finished
before embryosac development begins and so coincides with
megaspore-formation. When however there is only a row of
two (fig. 3 C) the second reduction division — and when there
is no row at all (fig. 3 D) both divisions — are shifted into
the germinating megaspore.

We have already moved some arguments in favour of our
treating these two processus separately (p. 15, IG and p. 19).
Moreover the correctness of this conception will be confirmed
later on, this idea of a reduction series being in full accordance
with the vacuolation process.

2°. The omission of celhoall formation and its consequences for
the number of nuclei in the mature embryosac.

Omission of cell walls during megaspore-formation is of rather
common occurrence in Angiosperms. It results in two, three
or four megaspores lying in the same cell. Of course this omis-
sion is not confined to those cases in which a quot;row of fourquot; is
formed (fig. 3 A). Megaspore-formation of the type B and C may
be\' affected as well. Fig. 4 presents a fully worked out scheme
of all further possibilities.

Cell- wall-formation is omitted in the second division only.
Thus two nuclei are met with in the same cell. Either one of
these may develop and the other desintegrate (fig. 4 ka) or both
of them may function (fig. 4 AA). The first should be in accor-
dance with the normal development, giving rise to an embryo-
sac of eight nuclei at most. The second however is still far
from improbable for both nuclei, lying in the same cell, have
got almost equal chances.

2®. Cell wall-formation is omitted in both divisions. Four me-
gaspores are found in the same cell. For reasons just mentioned
we might, expect either one of these to develop (fig. 4 Ay) or
all four (fig. 4 AAA A), which whould mean a really quot;tetraspo-
ricalquot; sac capable of becoming 32-nucleate, when fully developed.

For completeness\' sake two more possibilities are figured,
resp. showing two functioning (fig. 4 AA ft) and three functioning
(fig. 4 AAA a) megaspore nuclei. Their realisation however does
not seem very probable.

1°. Cell wall-formation is omitted in the second division only.
Two nuclei in the same cell, either one (fig. 4 B«) or both
(fig. 4 BB) developing. ^

2°. Cell wall-formation is omitted in both divisions. Three me-
gaspores should be found in the same cell. This however must
be considered utterly improbable, for it can hardly be expected.

four megaspores; two of

them not separated by Typo Aa

wals; only one develops

four megaspores; two of

them not separated by Typo AA

wals; both develop

•three megaspores; two of

them not separated by Typo Ba

walls; -only one develops

threo megaspoies; two of

them not- separated, by,Typc BB

walls; both develop

four inogasporos, not »c|ia-
rated by walls; one do- \'I\'yP® ^y
;Vclopjnbsp;(j,

four megaspores, not sepa-
rated by cell walls; two Type AA/3
develop

foiirnicgnspore«, not sepa-
aled by cell walls; three Type AAA«
develop .

rnbsp;■ i.quot; •

fuur megaspores, not sepa-
rated by cell walls, all TypoAAAA
four develop

quot; A threo nienasporos, not so-
O j pnraled by ccll walls; Typo BBB
o J nil ihrca develop

* two mcjlt;a»porc*, not fcpa- \'

rated by coll walls; only Typo Ca
one dovclups

two mcgsjporcs, nottcpa..
rated by ccll walls; both Typo CO
develop

Kig. l. Origin of ii-, {IrU) and teiratporical tact. Letters corresponding
to those. uicd in fig. 3. The number of capitals indicates the number
of »pores entering in cmbryosac fonnation. Tho greek letter indicates
\'»ow many dcsorganizing megaspores are met with within te samo tac.

that of two daughter nuclei, lying in the same cell and being
under the same conditions, only one should divide. Fig. 4 B /?,
BB a and BBB are only inserted to make the scheme cover all
possibilities.

One sole division, in which cellwall-formation is omitted,
both megaspores thus being involved in the same cell. Either
one (fig. i C a) or both (fig. 4 CC) may develop.

Though megaspore-formation has been described in very many
cases, it is still impossible to produce a more or less complete
list of the various types. First of all only comparatively few
records, according to Coulter and Chamberlain (1912, p. 7G),
can be accepted without reserve, even regarding the number
of megaspores formed! Secondly in almost all publications va-
cuolation is wholly left out of discussion, the figures being often
too scanty to allow any conclusion as to the megaspore-type.
Therefore all attempts at compiling a complete list have been
given up. The following artificial scheme however furnishes a
serviceable survey of all possibilities and of a few necessary
instances to illustrate them.

Ikforo piLSsiiig on to the study of tlie further development
of the embryosac, three observations must still be made.

Attention must be called to the system which underlies our
indicating the various types by the formules used in fig. 3
and fig. 4.

1nbsp;capital : monosporical sac

2nbsp;„nbsp;s: bisporical

3nbsp;„nbsp;s: trisporical

4nbsp;„nbsp;s: tetrasporical

quot;Further it\'must bequot; emphasized that our terms: mono-, bi-
and tetrasporieal are by no means identical with those of
Coulter\'s. Deduction has lead us to distinguish; quot; ^
nine types of monosp. sacs, viz. A, A«, ky, B, B«, (B/3), C, Qx, and D.
five types of bispor. sacs, viz. AA, (AA/S), BB, (BB;^), and CC.
(two types of trispor. sacs, viz. AAA^z, and BBB).
one type of a tetrasporic. sac, viz. A AAA.
Coulter however, identifying megaspore-formation and chroma-
tine-reduction, discerns three types only:
monosporical sacs (including our types A and B)
bisporical sacs (including our types C, AA and BB)
tetrasporieal sacs (including our types D, AAAA and CC).
This distinction is not a theoretical question of nomenclature,
but based on a real difference. Our 17 types are no fancies but
plainly distinguishable forms. For instance Lilium, Peperomia
and Gunnera are all three considered from Coulter\'s point of
view, to be of a tetrasporieal nature and of the same megaspore-
formation. In fact however the early stages of development are
not the same, vacuolation following the first division of the
embryosac-mothercell in Lilium, the second division in Gunnera
and the third division in Peperomia. Lilium therefore must be
classified as belonging to type D, Gunnera as to type CC and
Peperomia as to type AAAA (fig. 2, p. 22).

Thirdly it is good to point out again, that actually two lines
of development are joined in fig. 4quot;, viz. the omission of cell-
wall-formation and the germinating of two or more megaspores.
Of these two the first line is fully worked out; the other one
however only inasfar as it coincides with the first one and thus
influences the number of nuclei in the developing embryosac.
As already stated above, we have left out of discussion the
very many cases in which two or more normally formed me-
gaspores (separated by cell walls) are seen functioning.

Normally the primary micropylar nucleus by two successive
divisions gives rise to a group of four nuclei. Spindles in the
second division are almost always showing \'J\' shape. The upper

sister nuclei are the synergids, the other two being the egg
and the upper polar nucleus.nbsp;-nbsp;\' .

Often the development of this group is affected by a
reduction in the number of nuclei. Theoretically an increase
should be possible as well, but instances of a regular occurence
of more than four nuclei are not known.
■ The reduction series is worked out in figure 5, the number

of nuclei ranging from four to one. A further reduction should
not bo possible, for the character of the gametophyte resists
against total suppression and requires at least one nucleus:
the egg. When four nuclei are present, one is the egg, one
the upper polar nucleus and two arc synergids (fig. 5 1).
Three nuclei may be either the egg and two synergids (fig. 5

lla) or the egg, the upper polar and one synergid (fig. 5 lib).
Two nuclei are egg and synergid (fig. 5 llla) or egg and polar
(fig. 5 Illb). One nucleus necessarily must be the egg, being
either the undivided primary micropylar nucleus (fig. 5 IV) or
possibly even the undivided megaspore itself (fig. 5 V). A sup-
pression of the egg does not seem very probable. The case of
Dasylirion in which there should be no egg in the micropylar
group has been proved to be false. As long as no new cases are
reported, we can safely leave out of consideration such a possibility.

If we examine conditions more closely, it must be admitted,
that this scheme really combines two reduction processes. The
one first attacks the synergid-development, the other begins
with a suppression of the egg-polar divison. An exact illustration
of the situation is presented by fig. 6.

Fig, g. IJcJuctioii of the inicropylor groiip, c = egg s = syncrgiil
p = nppcr polar niicleii».

To confirm this conception about the development of the
micropylar group the following list of instances drawn from
literature is offered.

Type I : the quot;normal developmentquot;.
Type Ua : Aglaonema (OAMPnKi.i,, 1012), Harcinia (Tiifun,
1911), Moringa (Rutckhs, 1922), Cypripedium (Pack,
1907), Gastrodia (Kusano, 1915).
Type II/; : (Juglans regia (Kaiistkn, 1902)?)
Type Illrz: Peperomia (Johnson, 1900, 1907,1914, Brown, 1908)
(Dicraea elongata (Maonds, 1913)?)

Type IIW: Plumbagella (Dahlgren, 1915).
Type IV : Plumbago (Dahlgren, 1915).
Type V :......... .

The primary chalazal nucleus also, normally develops into a
group of four nuclei: the lower polar nucleus and three antipodals.

Figure 1 illustrates the theoretical reduction series. The
scheme covers all possibilities from the normal number of four

Fig. 8. Dcsiutrcgralinj? nuclei in the chalatal group. Letter* corrc»-
punding to those used in fig. 7. The Greek letter indicates the
number of desorgonizing nuclei.

nuclei down to total suppression of the entire group. Some-
times, when the upper polar nucleus is suppressed (fig. 5,

Ha, Ilia, IV, V) two of the chalazal nuclei are seen functioning
as polars and even fusing.

Besides the reduction just mentioned there is still another
way by which the number might decrease. Antipodals very
seldom survive the fertilization stage. Often however they begin
desintegrating long before the sac has reached its full-grown
stage. This of course is to be considered as an anticipation
without much interest from a morphological point of view. Foi*
completeness\' sake all possibilities on this line are reviewed
in figure 8. The same indicating letters are used as for the
corresponding embryosacs of fig. 7; a Greek letter is added to
indicate whether one, two or three nuclei are affected by early
degeneration.

The antipodals have not acquired special functions like the
nuclei of the egg-apparatus. Ordinarily the chalazal group is
formed by two simultaneous divisions. For these reasons a num-
ber of three chalazal nuclei is not very probable. As a matter
of fact no instances could be found of the types 2 b (fig. 7) and
2a;«, 2a/3 and 2hx (fig. 8).

Representatives of the various types are presented in a table
on the following page.

An increase in the number of antipodal nuclei is very com-
mon too. This increase, however, seems to be of secondary origin
for it is caused by a development of an original number
of three nuclei. Most probably it is connected with special nu-
tritive functions of the antipodal apparatus and is of no interest
from a phylogenetical point of view. A complete list of all
cases in which the number of antipodals surpasses the usual
number of three is given by Samuels (1912 p. 100).

As most authors have not paid much attention to the dis-
tribution of the protoplasm and to the arrangement of the nuclei
during the early stages of embryosac development, this review
necessarily [must be a[critical one, including not [only the inter-
pretation of the figures as given by the\\iuthors, but a discus-
sion of the figures themselves as well.

The common désintégration [of an otherwise normal]group jof
antipodals (fig. 8, la;«, Ma/3, la?\') as J well as the secondary
increase of the number] of antipodals, are not included. For the
rest the criticism reckons with all embryosacs, known to have
more or le.ss than the ordinary [number of eight ^nuclei, and it
thus covers the whole range of possibilities represented in fig. d—8.

Ppporoniia Sintonnii
Peperomia arifolia
Peperomia Ottomania
Peperomia rosediflora
Peperomia hlanda
Pci)eromia marmcirata
Peperomia iiiagnoliifolia

Euphorbia palustris
Acalypha
Garcinia Kydia
Garcinia Treubii
Moringa oleifera
Oenone Irathurni
Oenone guyanensis
Oenone Richardiana
Oenone Treslingiana
Oenone Versteegiana
Oenone marowynensis
Apinagia divertens
Apinagia Goejei
Apinagia perpusilla
Lophogyne capillacea
Mourera fluviatilis
Trislicha hypnoides
Rhyncholacis raacrocarpa
Oenone Ilulkiana
Cladopus Nymanni
Podostemon subulatum )
Ilydrobium olivaceum -
Farineria metzgerioidcs\'
Lawia zeylanica
Dicraea elongata
Sarcocolla squamosa
Sarcocolla fucata
Sarrocolla forniosa
Penaea mucronata
Penaea ovata
Ilrachysiphon imbrica-
tuni

5-, occas.
7-aucleate

IG-nucleate

16-nucieate

IG-nucleate
16-nucleate

5-nucleate

4-nucleate

5-nucleate
4-nuclcatc
4-nucleate

lO-nucleato

4-nucleate
\'i-nuclcatc

-i-nucleate
4-nucleate

4-nnclcatc

4-nucleate

Oenothera nutans
Oenotliera pjcnocarpa
Gaura Lindheinieri
Gaura parvifiora
Godetia spec.
Jussieua repens
Ludwigia prostrata
Circaea quadrisulcata

191G

1915
1910

191G

1914

1915

Ilroughtonia sanguinea
Coralliorrhiza innculata
Phajus grandillorus
lUotia Slicplicrdi
Gyrostachys gracilis
(iyrostacliys cernua
Gastrodia ehita

1902
1914
1911

1907

1911

1912

! i914
! 1915

Juglans regia according to Karsten (1902) shows a quite nor-
mal development of the chalazal group. At the other end how-
ever only three nuclei should be formed, viz. the egg, the upper
polar nucleus and one synergid, which should correspond to the
formule D—lib—la. As the publication dates from 1902 con-
firmation of this condition is wanted. The more since other
Juglandaceae are reported to be quite normal in this respect.
Moreover the author mentions some details which he could not
explain sufficiently.

First of all his statement that polars never fuse and even
are often found wide apart. Secondly the fact, that the two
nuclei at the micropylar end, which should represent the egg
and the synergid, show no difference in size or construction.
And thirdly his mentioning three cases of sacs in which, after
fertilization, three dividing nuclei were seen, both nuclei at the
top (synergid and egg!) still being undivided. The author\'s ex-
planation is, that the second male nucleus has fused with one
polar only. The figures should represent the second mitosis of
this fusion-nucleus and a first division of the other unfertilized
polar nucleus. A very doubtful hypothesis indeed.

Is it not safer to ascribe to Juglans regia a normal egg-appa-
ratus? Of course this is a suggestion only, which needs verifi-
cation by a renewed investigation. But Karsten\'s publication
itself seems to contain rather strong arguments in its favour.
It is stated that cellformation occurs very late in Juglandaceae,
in Juglans nigra even not before fertilization. In my opinion
the two «polars, never fusing and often wide apartquot; are no
polars, but the egg and the fusion nucleus, Juglans regia being
8-nucleate and quite normal except as to the cell-foriLtion of

the egg, which lies free in the sac cavity at least until fertili-
zation. The only figure published by Karsten most strongly
supports this suggestion, showing two absolutely equal cells (the
synergids!) at the top, while the lower of the two nuclei in
the sac cavity (the fusion nucleus!) is of about twice the size
of the upper one (the egg). It is hardly possible to apply to the
said figure Karsten\'s interpretation of one synergid and the egg
at the top and of two polars .in the sac. Moreover my sugges-
tion gives a reasonable explanation of the three dividing nuclei
seen after fertilization, these being the egg- and the endosperm-
nucleus resp. in first and second mitosis.

Helosis guyanensis (Chodat et Bernard, 1900) very evidently
claims the formula D —I—4a. The disintegration of the primary
chalazal nucleus is already to be seen at the two-nucleate stage
of the embryosac. As a rule the whole nucleus has disappeared
before the micropylar\'s first mitosis. Only once two chalazal
nuclei have been observed.

Peperomia peUucida was described by Campbell as long ago
asnbsp;and reinvestigated b}»- Johnson in 1900. Campbell was

quite sure about the sac being IG-nucleate but he did not suc-
ceed in tracing the further history of the sixteen nuclei after
their formation. This gap however is fully filled up by Johnson.
Up to the IG-nucleate stage both authors agree even in details.
Neither in the two-nucleate stage nor in the four-nucleate one
any sign of polarity or vacuolation is to be seen: quot;Die vier
Kerne sind gleicliniassig vertheiltquot;. Not before eight nuclei are
well established vacuolation commences. Very soon ji large cen-
tral vacuole is formed and tiie eight nuclei are fouiul periphe-
rically. A simultaneous division gives rise to the 16 nuclei of
the full-grown sac. Campbell supposed that afterwards three
nuclei should come together at the top of the embryosac, for-
ming the usual egg-apparatus, but he was not quite sure about
their always numbering three. (His fig. S on his Plate XXXI
reproduces only two nuclei). Johnson cleared up the matter
and there can be no doubt now that there is only one synergid
besides the egg. A similar group of two cells is found at
the chalazal end and two other groups lie lateral. The eight re-

There seems to be no difficulty at all in the interpretation
of the phenomena on the basis of our schemes. Vacuolation
does not begin before the eight-nucleate stage, so the four-
nucleate stage cannot represent anything else but four mega-
spores. Each of these four gives rise to a primary micropylar
and a primary chalazal nucleus, and further by a second divi-
sion to four nuclei two of which, belong to the micropylar and
two to the chalazal end. The two micropylar ones arrange
themselves as egg and synergid, the chalazal ones as two po-
lars. The mature embryosac thus contains four egg-apparatus
of two cells each, and eight polar nuclei, corresponding to the
formula AAAA—Ilia—3b (fig. 9, p. 43).

This conception is confirmed by Johnson\'s remark about the
synergid that quot;the position of the spindles in certain cases
seems to indicate that this is a sister to the eggquot; and further
by Brown\'s (1908) description of other Peperomia\'s, which all
show the same development. Moreover the tendency to the
reduction Ilia—3b among the Piperaceae is demonstrated by
Palm (1915) mentioning abnormal sacs of this type in the usually
eight-nucleate Piper subpeltatum.

Peperomia hiyndula has almost the same development. Accor-
ding to Johnson (1907) liowever, only the micropylar egg-appa-
ratus remains intact, while the otlier three quot;micropylar groups-
do not come to the formation of cell-walls, thus leaving all
nuclei free, which results in their fusing with the eight polars.
In the full-grown embryo-sac only the egg, one synergid and
one huge primary endosperm nucleus are left. This however does
not affect the AAAA—Ilia—31) character as shown by the
development.

Peperomia Sintemii and the other species described by Brown
(1908) and Hauskr (191G) all sliow tlie same development. It
is not necessary to repeat everything in detail. The tetrasporic
character is emphasized by the fact that in the first division
of the embryosac-mothercell of P. Sintensii a evane.scent wall
is formed, while quot;when the two nuclei divide into four, plates
are formed on both spindles.quot; Moreover in P. resedifiora and

Fig. 9. 16-nucleate emhryotac of Feperomia. Four megaspores, e.ich developing a micropylar and

a chalazal groap, both of two naclei only. Eight fusing nuelci.
Fig. 10. The 16-nucUate emhryotae of Enphorbiaeeae and Penaeaceae. Four megaspores, (gt;11 of them only

developing a micropylar group. Four fusing nuclei.
Fig. 11. The 16-nveleaie embtyotac of Gunnera. Two megaspore«, both fully developing. Seven fusing nuclei.
Fig. 12. Tie IG-nncleuU embryosac of P^ethrum parihevifolium. Two megaspores, both fully developing.

P. blanda quot;werden bei beiden Schritten der Meiosis Wände
gebildet von sehr unregelmäsziger Lage, die jedoch nach kurzer
Zeit wieder aufgelöst werden.\'-\' In this case also we can safely
accept the formula AAAA—Ilia—3b.

Arnoldi\'s publication on ihoi Euphorbiaceae is most annoying inits
lack of detail. Moreover there are absolutely no plates and the pu-
blished figures are too few in number to base any conclusions
upon. The author seems to have suffered from lack of material.
Ceramanthus mature embryosac probably contains only the

Ende sah ich keine Kerne, obgleich dies nicht für absolut
gewiss gelten kann.quot; The only figure of the two-nucleate stage
shows clearly a primary chalazal nucleus as well as the pri-
mary micropylar one. The next figure represents four nuclei
at the micropylar end. Evidently the chalazal nucleus degene-
rates very soon after its formation, which is expressed by the
formula A—I—4^. The life history of the Codiaeum is confined
to five lines and four figures. There seems to be a four-nucleate
sac at the end, but how it originated cannot be decided. Ar-
noldi\'s figures point to some development of the primary cha-
lazal nucleus. Possibly the sac-nucleus in full-grown state is
the result of a fusion of two polars, the sac thus being not
four- but actually five-nucleate, and the formula A—I-3a«.
Pedilanthus as a rule corresponds to A—I—4. Occasionally the
reduction has not gone so far, there being two antipodals
left: A-I-2a.

The IG-nucleate Euphorbiaceae, described by Mouilewski (1909,
1910, 1911) and by Arnoldi (1912) are of such remarkable uni-
formity as to the development of their embryosacs that it is
not necessary to treat them separately. Vacuolation does not
commence before the eight-nucleate stage, which proves a tetra-
sporic condition. Each of the four megaspores develops only a
quot;micropylar groupquot;; the chalazal groups are wholly suppressed
and not even a primary chalazal nucleus appears. The four
quot;micropylar groupsquot; are to be found: one at the micropylar
end of the sac, one at the chalazal end and one at each end
of a transverse axis. Each group organizes an egg-apparatus.

leaving one free polar-nucleus in the cavity of the sac. Ulti-
mately these.four polar nuclei fuse, giving rise to the primary
endosperm nucleus. The whole development thus answers to the
form. AAAA—I—5. (fig. 10, p. 43) which is in close agreement
wdth the condition in other Euphorbiaceae. Dessiatoffs (1911)
incredible statement about there being a full tetrad and still
a 16-nucleate sac in Euphorbia virgata could not be confirmed
by Modilewski (1911) as we can easely understand now.

Garcinia Kydia and G. Treubii as described by Tredb (1911)
and Moringa oleifera Lam. (IIdtgers 1922) show a reduction in
both groups. At the micropylar end the division, which ought
to give rise to the upper polar nucleus, does not occur, at the
other end development is stopped after the first division. These
two chalazal nuclei fuse and act as embryosac nucleus, the whole
development thus corresponding to A—TIa—3b.

The Podostcmaceae investigated by Went (1909, 1910, 1912)
are of a remarkable uniformity as to the development of the
embryosac. There is no doubt al)Out the existence of a primary
micropylar and a primary chalazal nucleus in the two-nucleate
stage. The chalazal nucleus soon desintegrates while the other
one develops normally. Formula C—I—4«. Lawia Zeylanka (Mag-
nus, 1913) is less reduced. Here the primaiy chalazal nucleus
fuses with the upper polar nucleus: C—I—4. Podostemon sidm-
lalns, Hydrohiiim olivaceum and Favmevia Metzgcvioides on the
other hand seem to represent a more reduced condition. The
full-grown embryosac shows the customary four nuclei, but in
these cases no desorganising or fusing nuclei are to be seen
during the development of the sac. It is evident that the cha-
lazal group is entirely suppressed, for the usual polarity and
vacuolation at the two-nucleate stage is missing. The diderence
between the two-nucleate stage of e. g. Lawia and that of Po-
dostemon etc. can be illustrated by a comparision of lig. 9
(Taf. XI) and fig. 50 (Taf. XIV) of Magnus\' publication. Also
the. direction of the spindles in the next (last) division leaves
no doul)t about the micropylar character of all four nuclei. It
is true that the four nuclei are not crowded together at the
top of the sac as may be seen by the Onagraceae, but the

extraordinarily small dimension makes a spreading of the
nuclei through the whole of the sac inevitable. There seems
to be no reason why the formula C—I—5 should not be used
for these Podostemaceae. Dicraea elongata presents another con-
dition. According to Magnus the mature sac consists of one
synergid, the egg, and two antipodals. Only two or three of the
earlier stages have been seen and a complete series could not
be given. Under such circumstances it is difficult to decide about
the formula. The two-nucleate stage is clearly polarised and vacu-
olated, and the next division would suggest also C—Ilia—3a.
This however differs widely from the other Podostemaceae.

The Penaeaceae as far as investigated (Stephens 1908, 1909)
show quite the same development as the Euphorbiaceae. The
arrangement of the nuclei and the organisation of the vacuoles
clearly shows a tetrasporic origin and suppression of all four
chalazal groups, the formula thus being AAAA—la—5 (fig. 10,
p. 43). The quot;micropylarquot; character of the lateral and chalazal
groups of cells is emphasized by the fact that embryos were
seen arising from one of these groups.

The Onagraceae, (Geeuts 1909, Modilewski 1909, Werner 1914,
Renner 1914, Tackholm 1914, 1915, Ishikawa 1918) show an
absolute uniformity in their development. No need to describe
the several stages in detail. The spreading of the protoplasm,
the vacuolation, the direction of the spindles and the crowding
of the four nuclei at the topend of the sac, these all make the
total suppression of the chalazal group so evident that Geerts
already said: quot;In der Oen. Lam. ist die erste Teilung im Em-
bryosack ausgefallen, und es entstehen somit gar keine Anti-
poden und kein unterer Polkern.quot; Form. A—I- 5. Ho however
did not recognise the individuality of the nuclei, for he still
homologised the two-nucleate stage with the same stage in
other Angiosperms, supposing the chalazal nucleus to be dis-
placed by protosplasm-stream.

Of several Gwmera species the life history of the female ga-
metophyte is published. We can pass the fii-st publication
(ScHNEGG, 1902) on the subject as his record could not be con-
firmed by Ernst (1908). All other publications, however, (Modi-

lewski lÔÔS, Ernst 1908, Samuels 1912) agree on most of the
important points. The first division of the embryosac-mother-
cell nucleus is not followed by wall formation and from the
figures it is clear that the large vacuole is formed at the four-
nucleate stage. From what is said in a previous chapter we
must conclude that the Gunnera sac reprents a bisporic con-
dition, polarisation leading us to homologize the four-nucleate
stage with the two-nucleate one of normal sacs. Both mega-
spores of the embryosac come to full development, thus giving
rise to two micropylar and two chalazal groups of 4 nuclei
each. In the mature sac one of the micropylar groups (the
egg apparatus) is to be^ found at the top, both chalazal groups
come together at the bottom, while the second micropylar
group, in which cell formation is omitted, fuses with the three
polar nuclei of the other groups. This agrees with the figures
published by the different authors, showing an ordinary egg-
apparatus, six antipodals and seven fusing nuclei. Special attention
might be called to the fact that this explanation of the Gun-
nera embryosac, based on the vacuolation, results in a reasonable
explanation of the puzzling number of seven fusing nuclei. The
development corresponds to CC—1—la (fig. 11, p. 43).

JHumbago species, investigated by Dahlgren (1915, 191G) all
showed a four-nucleate sac. The two-nucleate stage is clearly
polarised, both the primary micropylar and the primary cha-
lazal nucleus giving rise to a group of two nuclei, viz. an egg,
two polars and one antipodal, which corresponds to D-IIlb-3a.

Cemtost\'ujma (Dahjajren 1910) normally develops in the same
way. Occasionally however the antipodal degenerates. In Phm-
baijella (Daiii.grkn 1915, 1911\')) this désintégration is fixed, the
full-grown sac never containing more than three nuclei:

Pi/rethnim partheni/olium var. aurcum (Palm 1915) supplies
another instance of a bisporic sac. In the two-nucleate stage
plasm in still homogenou.s, vacuolation immediately following
the next division. As in Gunnera, here too, the bisporical cha-
I\'acter is accentuated by the direction of the spindles and by
the early stages of vacuolation. quot;Bei der zweiten Teilung im

Embryosack nehmen die Spindeln eine schiefe Stellung zur
Längsachse des Embryosackes einquot; states Palm. This deviation
from the ordinary condition and still more the vacuoles originating
in two distinct groups strongly influences the character of the
embryo sac. The figures themselves seem to suggest that there
are two developing megaspores.

Both megaspores develop their full number of eight nuclei.
Owing to the narrowness of the Pyrethrum sac the nuclei of
the different groups do not mix up. The sac thus presents a
row of four groups of four nuclei each, in fact a row of two
eight-nucleate embryosacs.

In the upper one the nuclei are arranged in the ordinary
way, there being two synergids, an egg, two fusing polar nuclei,
and three antipodals. By these three antipodal cells the narrow
sac is barricaded so that communication between the upper
and the lower half is made impossible. In the lower sac the
behaviour of the nuclei is somewhat abnormal. Four of them
separated by cell walls are to be seen just below the three
antipodals mentioned. The other four remain free in the sac
cavity. Perhaps, however, no mature sacs are seen by the
author, as Palm says: quot;lieber die spatere Entwickelung dieser
eigenartigen Zelle gibt leider mein Material keine sichere Aus-
kunft.quot; As far as present knowledge reaches the formula mu.st
be Cc—I—la. (fig. 12, p. 43).

Pyrethrum thus closely agrees with the related Tunacetum
(Palm 1915). In both the details of the development are the
same, the only difference being the somewhat further reduced
stage of the Tanacetum embryosac, in which only four nuclei
are developed by the lower megaspore. It is almost incredible
how Palm, who has been struck himself by this strong agree-
ment, could have been so fascinated by the idea of the quot;16-
nucleate typequot; that he, in spite of relationship and agreement,
separated Pyrethrum from Tanacetum and classified it in the
same gi-oup as Peperomia, Penaeceae, etc.

Limnocharis emarginata has been investigated twice. We can
pass Hall\'s publication (1902) on the subject, as his improbable
statement of the life-history was proved to be false by Nitzschke

(1014). This author describes the micropylar group as quite
normal, while the primary chalazal nucleus divides only once,
and sometimes twice, the six- or eight-nucleate sac thus cor-
responding to A—T—3a (or la).

Clintonia horealis embryosac (Smith 1911) is of the Ay—I—5
type. Here too the type is as pronounced as it was in the
Onagraceae, and the resemblance to the Oenothera-sac Smith
did not fail to notice. Perhaps the suppression of the chalazal
gi\'oup may have been induced in this case by the peculiar tetra-
sporical condition.

C^pripedium (Pace 1007) was one of the very first abnormal
sacs discovered, and has never been reinvestigated since. Though
the description is by the hand of an eminent examiner like
Miss Pace, when studying the figures of her richly illustrated
article the conclusion forced itself upon us that not only an-
other more probable interpretation is possible, but also that
Miss Pace\'s interpretation does not cover all the data furnished
by her illustrations. The author describes the development as
follows»: quot;The megaspore-mother-cell gives rise to two daughter-
cells, of which the inner one (exceptionally the outer one) be-
comes the embryosac. (fig. 3, type C, p. 25). The primary embryosac-
nucleus divides only twice. The direction of the spindle in the
first division is | , of the spindles in the second divisin J . In
the two-nucleate stage a large central vacuole between the
nuclei is to be seen. Later the four nuclei should arrange
themselves as an egg-apparatus, consisting of two synergids
and the egg, and one free nucleus at the bottom or halfway
the\' embryosac. At fertilisation triple fusion should occur be-
tween this free nucleus, one of the synergids and the second
male nucleus.

It is evident that a development like this cannot be explai-
ned by the thoughts which underlie this study, based as they
are upon a fargoing specialising of the nuclei. The large central
vacuole in the two-nucleate stjige places the character of these
nuclei as micropylar and chalazal beyond doubt, and so accor-
ding to Pace the chalazal group should have provided the egg,
to say nothing about a synergid acting as upper polar nucleus.

The remarkable waj^ in which this review of abnormal embryo-
sacs seems to establish our views, made us doubt Miss Pace\'s
description and look for another explanation of the data.
We will give first some facts, not in agreement with Miss
Pace\'s view, then our own suggestion about the life history of the
female gamethophyte of Cypripedium, followed by our arguments
taken from Pace. Of course we cannot give more than a sug-
gestion, a definite decision being only possible by reinvestigation
of the whole material.

r. The statement about there being no more than two divisions
is based on the entirely negative argument that no more divisions
have been seen, which is recognised by Miss Pace herself in
saying: quot;No evidence of another division was found, although
at least 300 slides with hundreds of ovules upon each were
examined for this peculiar stage. When the sac is ready for
fertilization, four nuclei are present, so that if other nuclei are
formed they are very ephemeralquot;. On her Plate XXIV fig. 24
she however reproduces an unfertilised, ytye-nucleate sac!

IHie direction of the spindle in the first, and (according to
Pace) only division of the micropylar nucleus is |, while the ordi-
nary direction of the spindle in the division, which gives rise
to the two synergids, is —. :Moreover one of the figures (Pack,
Plate XXIV fig. 20) shows the two synergids still united by
fibres in — direction!

8\'. The statement about the entering of a synergid in triple
fusion is very poorly illustrated. As a matter of fact though
quot;double fertilization was observed in hundreds of instancesquot;
the removal of the synergid-nucleus to the embryosac-nucleus
has not been seen even once. The sacs show either, when
still unfertilized, both synergids in their place at the top,
or after fertilization, one synergid destroyed and two nuclei
below the egg. It seems to us that Miss Pace not knowing how
to trace the origin of that second nucleus, by lack of other
nuclei came to the conclusion that it could be nothing else but
a synergid-nucleus. This is a mere hypothesis however, and
cannot be meant to be more than that; she herself with ample

roaterial at haud, only saying that there are quot;two nuclei below
the egg, and from the lines of cytoplasm one seems to be the
synergid which has moved to that positionquot;. (Pace, Fig. 42 on
Plate XXV should demonstrate these lines of cytoplasm, but in
other figures e. g. 43 and 51 nothing of the kind is to be seen).

4°. It is well-known that the entering of the pollentube
means the immediate destruction of the synergid. All figures of
just fertilized embryosacs, published by Pace, seem to furnish
proof of this occuring also in Cypripedium. On Plate XXVI,
Fig. 44 she figures a synergid in which the pollentube has just
entered; the nucleus of this synergid is evidently desintegrating.
Her Plate XXV, fig. 43, at a little later stage, shows the two
fusing nuclei at the bottom, while in the synergid there are two
male nuclei and a stained thing which must be the last remnant
of the synergidnucleus. Her Plate XXVI, fig. 45 represents
double fertilization, while in the upper half of the embryosac
the deeply stained remains of both synergids are to be seen.

Our own suggestion about the development runs as follows:
In the. two-nucleate stage there is a primary micropylar and a
primary chalazal nucleus (Pace, Plate XXIV, Fig. 24). Let us
Ibllow the history of the micropylar one lirst. The direction of
the spindle in the division of this nucleus is j (Pace, Plate
XXlV, Fig. 25). But this is not the only division its Pace
supposed. It is followed by a division of the upper daughter
nucleus only, giving rise to the two synergid nuclei. The other
Jiucleus remains undivided and becomes the egg (Pace, Plate
XXIV, Fig. 20). As to the chalazal group only one division
«f the primary chalazal nucleus occurs. As a rule the prim,
dial, nucleus remains undivided until just before fertilization
(1\'ace, Plate XXIV, Fig. 20, Plate XXV, Fig. 29, 30), thus pre-
senting a really four-nucleate embryosac. Then it divides, the
two daughter-nuclei staying close together (Pace, Plate XXV,
l^\'ig. 42, 43, etc.) and acting as embryosac-nucleus. At fertili-
zation the pollentube enters one of the synergids (Pack, Plate
XXVI, fig. The nucleus of this synergid at once begins to
Regenerate (Pack, Plate XXVF, Fig. 44). A few moments later
^wo male nuclei have entered the synergid, whose nucleus is

irapidly shrinking (Pace, Plate XXV, Fig. 43). By the time of
the fusing of the male nuclei with the egg and the double
sac-nucleus, only very small remains of both synergids are still
to be seen (Pace, Plate XXVI, Fig. 45). While this seems to
be the ordinary course, sometimes the division of the primary
chalazal nucleus may occur a little bit earlier, even at the
same time of the division of the primary micropylar nucleus
(Pace, Plate XXIV, Fig. 25; Plate XXV, Fig. 27; Plate XXVI,
Fig. 46). Probably in these cases the fusion of both nuclei
occurs already before fertilization, leaving a fusion-nucleus
instead of the usual double-nucleus, as may be derived from
Pace\'s statement that quot;one sac indicated the possibility that
the synergid may fail to unite in the triple fusion,quot; or in
other words, that there was only one nucleus to fuse with
the second male nucleus. If this record of Cypripedium proves
to be right, the development of the embryosac corresponds to
the formula G—Ila—3b.

Our arguments for this interpretation can be summarized as
follows: 1. The insufficiency of the interpretation of Miss Pace
to declare all figures given by her. 2. The details of her figures
as indicated in my description. 3. The lack of evidence put
forward in her arguments for the entering of a synergid-nucleus
in triple fusion and for the chalazal \'origin of the egg, which
are both entirely without analogies. 4. The analogies presented
by Gastrodia, in which both chalazal nuclei fuse soon after
their formation, by Garcinia, in which they fuse just before
fertilization and by Moringa, in which fusion takes place after
fertilization, all three showing the same trinucleate condition
at the micropylar end.

As already stated we do not claim to have given a deci.sion,
this being impossible without reinvestigating the whole material.
But our suggestion must be admitted iis a possible explanation
and must be rejected on firm grounds before we can accept
Miss Pace\'s.

EpipactU piihesceiis, described by Biiown and Sharp (1011),
normally has the ordinary eight-nucleate embryosac. Sometimes
however the chalazal development stops at the bi-nucleate stage,

there bemg one polar and one antipodal nucleus. As all con-
ditions between full tetrade and embryosac-mothercell = embry-
osac, occur, the formula is: B (or A, C or D)—I —la (or 3a).

Broughtonia sanguinea, CoraUiorrhiza maculata and Phajus
gvandijlorus according to Sharp (1912) have undergone some re-
duction in the chalazal group by omitting the last division.
The only two chalazal nuclei fuse and act as lower polar
nucleus. Formula B—I—3b.

Bletia Shepherdi, investigated by the same author is either
normally eight-nucleate or reduced like the foregoing. Most
sacs, however, showed only four nuclei, more or less fusing or
desintegrating. Evidently this phenomenon can be brought back
to artificial growing conditions, as the author himself has
remarked.

Gyrostachys (Spiranthcs) gracilis (Pace, 1914) exceptionally
develops a normal eight-nucleate embryosac. Usually however
the primary chalazal nucleus divides only once, giving rise to
a polar nucleus and one antipodal which sometimes st^iys, but
mostly degenerates. Formula = B (or C or D)—I--3a;: (or 3a or la).
Gyrostachys (Spiranthes) cernua has quite the same development,
but occasionally a row of four megaspores is seen. Miss Pace
referring to her figures 34 and 35 assumes the eml.n-yosac so-
metimes to be four-nucleate. Whether these figures really
represent four-nucleate sacs might be doubted; the dimension
of the sac nucleus at least points to its being a fusion nucleus.
There seems to be no reason why these sacs should not be
originally six-iuicleate like the others. Pace\'s figures 31 and 30
show sacs with the antipodal gradually disintegrating, which
must result in sacs like 34 and 35.

Gastrodia elata is described by Kusano (1915) as four-nucleate
like Cypripedium; the primary micropylar nucleus should produce
two synergids, the chalazal nucleus the egg and the embryosac
nucleus. Here too a synergid should enter in triple fusion. It
is not necessary to repeat all that has been siiid gt;vhen criti-
cising Pace on Cypripedium. Everyone of my remarks holds
for Gastrodia too. Especially the fact that Kusano also ab-
solutely failed to see the migration of a chalazal nucleus to

the position of the egg (his figures 93—95 which should illustrate
this migration do not even give the slightest indication of a
chalazal origin of the egg!) and that he too absolutely failed
to see the migration of a synergid-nucleus towards the sac-
nucleus, might almost be called a proof of the incorrectness ot
the suppossed development of Gastrodia and Cypripedium.
Otherwise either Pace or Kusano should have found one or more
of the lacking stadia. Moreover Kusano\'s figures so strongly
resemble those of Tkeub (1911) on Garcinia and so strongly
suggest an explanation in that direction, that it is hard to
understand why he did not come to it.

After the elaborate criticism on Cypripedium I can do with
marking only a few of the most obvious phenomena in Gas-
trodia. First of all the fibres between the egg and synergids
as illustrated in Kusano\'s fig. 93 and 94, which seem to represent
the ordinary behaviour and are noticed by Kusano himself who
tried to explain them by saying: quot;Later, the limiting plasmic
membrane is precipitated between each two nuclei, often pre-
ceded by the formation of fibres.quot; To me it seems more reaso-
nable to accept a micropylar origin of the egg and to do
without the fiir-fetched explanation of the fibres. A second
phenomenon, which makes a chalazal origin of the egg not
only improbable but quite impossible is illustrated in fig. 88,
89, 90 and 91. Though Kusano himself says quot;it is almost cu-
stomary that they (viz. the chalazal nuclei) lie in close contact
(fig. 91),quot; he does not hesitate to consider the majority of his
material as abnormal! According to him the growth and divi-
sion of the chalazal nucleus should be much disturbed by the
lesser amount of cytoplasm, and all those sacs shoulds be
unable to come to full development. I do not think such a
presumption can be accepted unless every other possibility is
at least tested and rejected on firm grounds.

As far as can be gathered from Kusano\'s publication the
development seems to be as follows: The embryosac (one of a
row of three megaspores) is in its bi-nucleate stage clearly
polarized (Kusano, fig. 78, 79, 81, 86, etc.). The primary mi-
cropylar nucleus presents a reduced development, giving rise

to two synergids and the egg (Kusano, figg. 93, 94). The divi-
sion which should give rise to the egg and upper polar nucleus
is suppressed, there thus being no upper polar nucleus. The
primary chalazal nucleus on the other hand divides only once.
As a rule both nuclei fuse soon after their formation (Kusano,
fig. 89, 90, 91). Sometimes however this fusion may be a little
bit postponed (Kusano fig.85). Only two of Kusano\'s figures do
not agree with this suggestion. Both figures (80 and 84) represent
nuclear divisions which by the direction of their spindles and
by the distribution of the plasm point to sacs of the type
B—1—5, instead of B—Ila—3b. I must emphasize here that
of course it is not possible to give a description of a life history
without any material at hand. Only a thorough reinvestigation
can clear up the matter.

Aglaonema commutatim has been studied by CAMrBKLi. (1900,
1903, 1912). He did not succeed, how^ever, to give an idea of
the development. All his material of this species was collected
from plants grown under more or less artificial conditions, and
shows the common pathological phenomena like indefinite number
of nuclei, multiple fusions and other abnormalities. Moreover
quot;there is some evidence that the complete embryosac may be
the product of the union of several sporogeiious cells (inega-
spores).quot;

Ncphthjlis Uberica is not better known (CAMrnKU,, 1905). It
Avas „quite impossible to make out any prevailing typequot;, and
as this material too came from the greenhouse, the author
rightly remarks: quot;How far these are normal cannot be cer-
tainly determined until material grown under natural condi-
tions can bo exiimined.quot; More stress is still laid on the pre-
sumption of pathological conditions by .the fact that \'\'the pollen
grains were badly shrunken and distorted, and no satisfactory

Aglaonema pidum (Camphkll 1903, 1912) was lirst assumed
to luive a normal eight-nucleate embryosac. In his later publi-
cation CAMi\'nELL however describes the sacs as live-nucleate,

There should be a normal development of the micropylar group,
while at the other end a different number of nuclei, varying
from 2 till 11, should be produced. Gow however has not seen
all stadia of development. His record of what he has seen is
very brief and the figures very few in number. Moreover these
figures are only outlined and of no use for further research.

Aglaonema simplex and A. modesium (Campbell, 1912) are five-
nucleate and of the D—Ha—3b type.

Our knowledge of the embryosac development of most families
is yet too scanty to justify any attempt at an exact system of
the female gametophyte. The survey however still indicates
certain tendencies in development and it furnishes some unex-
pected evidences of relationship worth special mention.

Among Piperaceae Peperomia shows a regular development
of four megaspores and a regular reduction in the number ot
nuclei produced by each of these megaspores. Exactly the same
tendencies are occasionally met with in Piper. It is only by
analising gametophytes as we did, that this relationship between
a 10- and 5-nucleate sac came to light.

Euphorhiaceae may be either 10-, 8-, 7-, 5- or 4-nucleate. Super-
ficially any connection between these types seems to be lacking.
In fact, however, they are as closely related as possible. Only
the chalazal group of nuclei is affected by a process of reduc-
tion, which can go as far as total suppression. In the 10-nucleate
sacs this process is combined with tlie development of all four
megaspores.

In Penaeaceae the same combination of total suppression of
the chalazal group with a tetrasporicjil condition is to be seen.

All Onagraceae are like the Penaeaceac, except their developin^r
only one of the four megaspores.

Monocotyledones show a great variety of types. Reduction
processes are still going on in every direction. The number of
megaspores ranges from four to one, and both the micropylar

and the chalazal group may suffer from a decrease in the
number of nuclei. Among Orchidaceae these divergent lines of
development are even met with in the same family. In some
of the species the micropylar group is still quite normal, the
chalazal one being affected by reduction in the number of
nuclei; in other species the micropylar end too never reaches
the four-nucleate stage.

Our method of treating the various sections (megaspore-
formation — micropylar group — chalazal group) of the
gametophyte as morphological units, capable of following in-
dependent lines of development, proved to be a real progress,
in so far at least as closely related species are no longer scat-
tered all over a system. In this respect our outlines show
distinct advantages on Coulter\'s, Ernst\'s, Dahlgren\'s and Ishi-
kawa\'s schemes.

Especially the application of the idea to the 16-nucleate em-
bryosacs has been fruitful. These sacs are no longer a type of
their own, but are considered as either of bi- or of tetraspori-
cal origin. A fully developed tetrasporical sac should be 32-
nucleate, each megaspore developing a quot;micropylarquot; and a
quot;chalazalquot; group. Both however can be subject to reduction
conformable to the possibilities, worked out and illustrated in
figures 5—7. Several of these reduction types can be 16-nucleate.
Of course full development of a bisporical sac also leads to
a 16-nucleate embryosac.

Up to the present moment the Angiosperm embryosac has
held a wholly isolated position. Its origin could not be traced
and its analogies among Gymnosperms were dark. A gap in
phylogenetical knowledge on such an important point, necessa-
rily has led to numerous suggestions. None of these, however,
passed the hypothesis stage. It apparently depends in tiie main
on the author\'s preference for a certain theory on the origin
of Angiosperms, which type of embryosac he will call the most
primitive one. It must be admitted that this is not the right
way of settling the question, using phylogenetical speculations

The present study did not succeed in throwing more light
on the origin of the Angiosperm sac. It has led us to the view
that all embryosacs with an abnormal number of nuclei are
derived from the normal eight-nucleate type. This conception
of the 8-nucleate sac as the most primitive one fully agrees
with the actual conditions, for it is met with in all families
at the bottom of the natural system. On this point the phylo-
genetical value of our results is purely negative: it leads us
to reject all theories on the origin of Angiosperms, which are
founded on the quot;primitive characterquot; of the 16-nucleate em-
bryosac or of the embryosac with an increased number of
antipodals.

The reduction in the number of megaspores (quot;row of fourquot;,
quot;row of threequot;, quot;row of twoquot;, quot;no row at allquot;), in the number
of micropylar nuclei (four, three, two or one nucleus), in the
number of chalazal nuclei (four, three, two, one or no nucleus),
they all succesfully can be brought back to the same causes.

First of all the process of shortening the sex generation must
be mentioned. In most Angiosperms the sporogenous tissue has
been reduced to one cell only. So the next step on this way
necessarily must affect megaspore-formation and embryosac-
development.

Secondly there is the usual desintegrating and final suppres-
sion of non-functioning tissues. This is too well-known from
sporophytic conditions to need any further commentary. Its
application to the megaspore-formation and embryosac-develop-
ment will meet no objections.

The reduction in the number of megaspores probably is caused
by both processes. Normally three of the spores are seen des-
integrating. A total suppression should be an anticipation on
this degeneration.

both factors. Most authors who have made special study of the
subject agree on ascribing to the antipodals a nutritive func-
tion (Westermaier 1892, Ikeda 1902, Lotsciier 1905, Huss 1906).
Their losing this function leads to désintégration and finally to
total omission. This, of course does not hold for the lower polar
nucleus. Still this nucleus may be suppressed too, which shows
that the process of shortening the sex-generation is at work
as well.

At the niicropjdar end the nuclei, once formed, usually per-
sist. They have all got a special function. The occasional sui)-
pression of one or two of these nuclei therefore must be con-
sidered as a result of the shortening of the n-generation. This
conception explains why reduction in the number of megaspores
and in the number of chalazal nuclei is more common than in
the number of micropylar nuclei. The development of the mi-
cropylar group is affected by one reducing factor only, while
in megaspore-formation and chalazal development two factors
are at work.

An increase in the number of chalazal nuclei, on the other
hand, safely can be ascribed to a more intensive nutritive func-
tion. This view is supported by Campbell\'s (1899 a, 1899 b) state-
ment about the occasional increase after fertilization and in
relation to the nourishing of the embryo.

We have to mention here some literature about embryosacs,
showing special anomalies not in keeping with the results ol
the present study.

First of all Campbell\'s publications on Aglaonema commuta-
tum (1900, 1903, 1912) and on Nephthytis liberica (1905). We
have already cited these cases in our general review. Tho irre-
gularities are doubtless pathological and caused by the abnor-
mal conditions under which the material was grown (in green-
houses). Especially the many multiple fusions strongly remind us
of Nemec\'s studies on tho influence of external circumstances
on nuclear division and nuclear fusions.

Secondly three cases of two micropylar eggs are reported.
(Strasburger 1878, Fischer 1880, Murbeck 1902). In itself a
secondary increase in the mimber of micropylar nuclei is not
in opposition to our views. In more recent literature however
no such cases are met with. Therefore these records were not
inserted in our general survey, as they can only be accepted
under reserve of further confirmation. Especially since these
cases do not represent normal conditions, but anomalies. Fischer
himself even doubts the correctness of his observation, the only
indication being one section „dessen Tauglichkeit durch den
Schnitt leider herabgesetzt worden ist.quot;

Thirdly in a few publications a synergid is mentioned as having
assumed the function of an egg or of an upper polar nucleus.
Almost all of these studies are dated long before triple fusion
was known. The only exceptions are Pace on Cypripedium and
Kusano on Gastrodia, but their figures probably have been mis-
interpreted, as we have already discussed (p. 49—52, 53—55).

Lastly we have to deal with four cases in which there should
be an egg of chalazal origin (Chamberlain 1895, Tretjakow 1895,
Pace 1907 and Kusano 1915). There is no need to repeat again
whath has been said in our discussion on Cypripedium (Pace)
and on Gastrodia (Kusano). As to the other two: Marie Opper-
mann (1904), when reinvestigating the Aster embryosac, says
that «there was notliing to indicate the presence of an antipodal
eggquot;. She too noticed that often one of the antipodals becomes
larger than the other two, «but in no instance was I able to
find in this lowest cell an antipodal oosphere as described by
Chamberlain (1895)«. The embryosac of Allium odorum (Tretja-
kow, 1895) is of the normal 8-nucleate type. The author speaks
of embryo formation «zuweilen sogar aus alien drei Antipoden
und zwar ohne Befruchtung.« This development begins quot;erst
nach der Befruchtung der Eizelle«. This really seems to be

^ We seem justified in finishing this study by stating that all
literature on the Angiosperm embryosac confirms our views,
the only exceptions being: a few publications of too old a date

to be accepted without further confirmation, and two more
recent studies (Pace, 1907 and Kusano, 1915) in which however
figures are probably misinterpreted.

1.nbsp;Several attempts have been made to classify the various
types of Angiosperm embryosacs. These systems are based
either on the number of divisions between embryosac-
mothercell and egg (Coulter) or on the number of nuclei in
the full-grown sac (Ernst). They are wholly artificial and
therefore without any phylogenetical value.

2.nbsp;The female gametophyte is no morphological unit, but a
complex, as well as the sporophyte. A natural system there-
fore presupposes thorough and detailed knowledge of mor-
phology. It has to reckon with the following processes as
probably independent lines of development:

3.nbsp;Chromatine reduction usually accompanies the first divisions
of the embryosac-mothercell. Sometimes (in apogamous spe-
cies) it is omitted, which proves that it is not identical
with megaspore-formation.

Polarimlion is a function of the developing megaspore
(embryosac). It does not accompany megaspore-formation,
but megaspore-development. It commences as soon as me-
gaspore-development begins. It therefore provides us with
means of recognising megaspores, even when two or four
megaspores are lying in the same cell: as long as plasm
remains iiomogenous spore-formation is still going on, as
soon however as polarisation (vacuolation) commences we have
to do with germinating spores. Moreover a large central
vacuole enables us to tell the nuclei of the chalazal group
from tiiose of the micropylar group.

Megaspore formation usually leads to a quot;row of four/
Occasionally however only three or two megaspores are
formed, or even the embryosac-mothercell itself is seen
functioning as a megaspore. The omission of cell walls
during spore formation may affect the number of nuclei in
the mature embryosac: four germinating megaspores in
the same cell quot; give rise to a tetrasporical, three to a tri-
sporical and two to a bisporical sac. We are forced to admit
theoretically: 9 types of monosporical, 5 of bisporical, 2 of
trisporical and 1 of tetrasporical sacs. The further develop-
ment of the megaspores must be considered as wholly in-
dependent from their formation.

Development of the micropylar group of nuclei. Normally it
results in a group of four nuclei. Sometimes however the
number has been reduced. Theoretically this reduction can
go down to there being only one nucleus left, the primary
micropylar nucleus or even the megaspore itself thus assuming
the egg-function.

Development of the chalazal group of nuclei. Here all stages
of the reduction series may be met with, from the usual
four down to total suppression of the entire group. Even
when nuclei are still formed their désintégration is of
common occurence. On the other hand a secondary increase
in the number of nuclei sometimes has been reported.

4.nbsp;A few, most probably incorrect, records excepted, all publi-
cations on abnormal embryosacs seem to confirm this con-
ception of considering the Angiosperm embryosac as a
morphological complex.

5.nbsp;Another confirmation is to be found in its systematical
value. Relationships are established by the study of the
gametophyte as well as by the study of the sporophyte. Espe-
cially the application of our views to 16-nucleate sacs has
been fruitful. They are no longer a type of their own, but
of bi- or tetrasporical origin, each one of their megaspores
being open to the deviations, which we have worked out
with regard to the monosporical sac.

pylar- and chalazal development occasionally are affected,
can be traced back to two causes, viz. the shortening of the
sex-generation and the usual désintégration and final sup-
pression of non-functioning tissues. Megaspore-formation and
chalazal development are attacked by both factors, micro-
pylar development by the first one only.

7. The 8-nucleate sac seems to be the most original type of
the Angiosperm embryosac. The present study did not suc-
ceed in throwing^ any light on its origin.

l\'Acad. Imp. d. Sc. St. Petersb. IX.
Brown, W. H. (1908). The nature of the embryosac of Peperomia, Bot. Gaz. 46.

Brown, W. H. and L. W. Sharp. (1911). The embryosac of Epipactis, Bot. Gaz. 52.
Campbell, D. H. (1899a). Die Entw. des Embryo sackes von Peperomia pellucida
Kunth, Ber. d. D. Bot. Ges. 17.

--, (1902). Recent invest, upon the embr.sac of Angiosperms, Amer. Naturalist 36,

(Avena fatua L.), Proc. Calif. Acad. Sc. 3d scr. Bot. Vol. I.
CiiAMiiEHLAiN. Cit. J. (1895). The embryo sac of Aster Novae-Angliae, Bot. Gaz. 20.
, (1905). Alternation of generations in animals from a botanical standpoint.
Bot. Gaz. 39.

CitoDAT R. et Bernard C. (1900). Sur le sac embr. de rilelosis guyanensis, Journ.
d. Bot. 14.

Coulter, J. M. (1908). Relation of Megaspores to Embryosacs in Angiosperms,
Bot. Gaz. 45.nbsp;015

Coulter J M. and Chamrerlain, Cm. J. (1912). Morphology of Angiosperms,
New \\ork 1912.nbsp;o i 1

Dahlgren, K. V. 0. (1915). Der Embr. sack von Pluml.agella, Arkiv. f. Botanik 14

Plumbaginales, Kungl. Sv. Vetensk. akad. Ilandl. 56.
Davis, (1905). Studies on the plant cell, Amer. Naturalist 38

Dessfatofk, N. (1911). Zur Entw. des Embr. sackes von Eup\'horbia virgata W. R.,
Ber. d. D. Bot. Ges. 29.nbsp;«61

Ernst, A. (1902). Chromos, rod., Entw. dos Embr. sackcs und Befr. bei Paris
quadrifolia L. und Trillium grandinorum Salisb., Flora 91.

-, (1908a). Zur Phylog. des Embr. sackes der Angiosj). Ber. d. D. Bot Gcs \'gt;G(J

Verh. d. Schweiz, naturforsch. Ges., 91 Jahres-Vers., Glurus 1908 Band I
I\'isciiER, A. (1880). Zur Kenntniss der Embryosackentwicklung einiger Angio-
spermen, Jenaische Zeitschr. 14.nbsp;« o o

Geerts, J, M. (lOOO). Bcitr. zur Kenntn. der Cytol. uiid der part. Sterilität von

Oenothera Lainarckiana, Ree. d. Trav. Dot. Neerl. 5.
Gow, J. E. (1908). Studies in Araceae, Bot. Gaz. 40.

Hall, J. G. (1902). An emhryological study of Limnocharis emarginata, Bot. Gaz. 33.
HausER, Ii. (1916). Untersuchungen an Makrogametophyten von Piperaceen, Beitr.
zur Allg. Botanik, 1.

Huss, II. A. (1906). Beitr. zur Morph, und Physiol, der Antii)oden, Beih. z. Bot.
Centr. hi. 20.

•Iked.v, (1902j. Studies in (he physiol. functions of Antipodals, Bull. Coll. Agric.
Imp. Univ. Tokyo 5.

Lshikawa, M. (1918). Studies on the Embr. sac and Fertilization in Oenothera,
Ann. of Bot. 32.

Jacobsson-Stl\\sny, Emma. (1916). Fragen vergleichender Embryologie der Pllauzeu,

Sitz. her. Kais. Ak. d. Wiss. Wien, Abt. I, Bd. 125.
Johnson, D. S. (1900). On the endosperm and embryo of Peperomia pcllucida.
Bot. Gaz. 30.

seed-developm. of Peperomia hispidula, Amer. Journ. of Bot. I.
Karsten, G. (1902). Ueber die Entw. der weibl. Bliithen bei einigen Juglandaceen,
Flora 90.

Coll. of Agric. Imp. Univ. Tokyo 6.
Lötsciieii, p. K. (1905). Ueber den Bau und die Funktion der Antipoden in der

Angiosperm.-Samenanlage, Flora 94.
Magnu.s, Wehnkr. (1913). Die atypische Embr. Entw. der Podostemaceen, Flora 105.
Mc. Alliäter, f. (1909). The developm. of the embr.sac of Smilacina stellata.
Bot. Gaz. 48.

Modilkwski, J. (1908). Zur Embryohildung von Gunnera chilensis, Üor. d. 1).
Bot. Ges. 26rr.

--, (1909J)). Zur Embryobildung von einigen Onagraceen, Der. d. 1). Bot. Ges. 27.

und anderen Euphorbiaceen, Ber. d. I). Dot. Ges. 29.
Murdeck, Sv. (1901). rarthenogen. Embryobildung in lier Gattung Alchemilla,
, Lunds Univ. Arsskr. 30, 2, No. 7.
--, (1902). Ueber die Anomalien im Baue des Nucellus und des Emhrvosackes

bei jiarthenogen. Arten der Gattung Alchemilla, I-unds Univ. Arsskr. 38.
NEmec, D. (1910). Das Problem der Defruchtungsvorgänge und andere zytol. Fragen,
Berlin 1910.

Nitzsciike, Jon. (1914). Beitr. zur Pliylog. der Monokotylen, gegründet auf der
Embr.sackentw. apokarp. Nymphaeaceen und Ilelobien, Beitr. z. Biol. der
Pflanzen 12.

OlTERMAN, .Marie (1904) A contribution to the lifo history of Aster, Bot. Gaz. 37.
Dace, Lui.a. (1907). Fertilization in Cypripedium, Bot. Gaz. 44.

Renner, 0. (1914). Befr. und Embr. bild. bei Oenothera Lam. und einigen verw.
Arten, Flora 107.

Jard. Bot. Buitenzorg 32.
Samuels, J. A. (1912). Etudes sur le développement du sac embr. et sur la fécon-
dation du Gunnera macrophylla Bl., Arch. f. Zellforschung 8.
Sciinegg, Hans (1902). Beitr. zur Kenntniss der Gattung Gunnera, Flora 90.
Sciiniewind-Thies. J. (1901). Die Redukt. der Chromos, zahl und die ihr folg.

Kernteilungen in den Embr. s. mutterzellen der Angiosp., Jena 1901.
Schürhoff, B. N. (1919). Zur Phylogenie des Angiospermen Embryosackes, Ber. d.

Smith, R. Wilson (1911). The tetranucleate embryosac of Clintonia, Bot. Gaz. 52.
Stephens, E. L. (1908). A prelim, note on the embryosac of certain Penaeaccae,
Ann. of Bot. 22.

--, (1909amp;). Recent progress in the study of tlie embr. sac of the Angiosi).,

New Phytol. 8.
Strasburger, Ed. (1878). Ueber Befr. und Zelltheilung.
TäCKHOLM, G. (1914). Svensk Bot. Tidsskr. 8.

Tretjakow, S. (1895). Die Betheiligung der Antipoden in Fallen der Polyeinbryonic

bei Allium odorum L., Ber. d. D. Bot. Ges. 1.3.
Treub, m. (1911). Le sac embr. et l\'embryon dans les Angiospcrme.s, Ann. d. Jard.
Bot. Buitenzorg 24.

Went, F. A. F. C. (1908). The development of the ovule, embryosac and egg in
Podostemaccae, Ree. Trav. Bot. Néerl. 5.

---, (19J2). Unters, über Podostemaceae II, Verh, Kon. Akad. Wet. Amsterdam 17.

Werner, Elisabeth (1914). Zur Oekologic atypischer Samenanlagen, Beih. z. Bot.
Centr. bl. 32.

Westermaier, M. (1892). Zur Embryologie der Plianerogamen, insbesondere über
die sogen. Antipoden, Nova Acta Acad. Leopold. Carol. 57.

Voor een „natuurlijk eindequot; van leven bestaan geen bewijzen.
Veeleer moet een physiologische onsterfelijkheid worden aan-
genomen.

Aan Morgan\'s crossing-over theorie en zijn opvatting van een
groepsgewijze lineair gebonden zijn der dragers van de erfelijke
eigenschappen, kan slechts de waarde van een werkhypothese
worden toegekend.

De oorsprong van Zea Mays is niet te zoeken in Zea tunicata,
maar in een bastaard tusschen Euchlaena en een der Andro-
pogoneae.

De exostichos en eudostichos, die I3olk l)ij Zoogdieren beschrijft,
zijn niet homoloog met de stichi der Reptielen, volgens de delinitie
van VVokrueman.

De Fommi\'niferen vormen geen natuurlijke groep. Het is beter
(Ie Monothalamia met de Amoebozoa tot éón orde te vereenigen
en de Polythalamia als een afzonderlijke orde te beschouwen.

De opvatting, dat de löss een IJstijdvorming is, staat sterker
(liin de meening, welke hmir een interglacialen ouderdom toe-
schrijft.

, \'V-

sjti\'

V/ .S .nbsp;■■ f\'Jr-

	er re
A il	3
quot;g i i	CL.


0 .

Type A
the quot;normal
typequot;
Type Act	Type A A
Calopogon	Smilacina
(i\'ack 1909)	^mc. au.istku 1909, 1914)
Type Ay	Type AAfi	Type AAA
Clintonia
(Smith 1911)
Avena (Can-
non 1900)
-------------------	----- - . .

■ ^ 3	Tj\'pe la:	Type la«-laj3-lay:
o a	the quot;normal developmentquot;	quite common
p — r* ci	Type lb:	Type Iba-lb^:
tc 2 quot;o
«1	Type 2a:	Type 2a«-2a^:
o c	Pedilanthus (Arxoi.di -1912)	............
CI, quot;5 ^ N p -2	Type 2b:	Type 2b«:
SD M u
	Type 3a:	Type 3a«:
	Dicraoa (Magnus 1913)?	Codiaeum (Arnoi.di 1912)
	Limnocharis (Nitzsciike 1914)	Plumbagelia (Daiii,gren 1915)
	Gyrostachys (Pace 1914)	Gyrostachys (Pace 1914)
quot;o	Epipactis (Brown amp; Sharp 1911)
c	Type 3b:
	Peperomia (camrneix 1899,
N r!	Johnson 1900, 1907, 1914,	•
quot;rt	Brown 1908).
	Garcinia (TuEun 19il)
^ 0	Moringa (Rutgers 1922)
1 ^	I3ronghtonia and other Orcliidac
1 3	(Shari- 1912)	I
5)	Cypripcdium (Pack 1907) Gastrodia (Kusano 1915) Aglaonema (Camimikm, 1903, 1912)	I \' i
_ ^	1 Type 4: j	Typo 4ai*:
\' i ®	Lawia (Magnus 1913) t	Ilclosis (Ciiodat amp; HERNAni) 1900)
i 1 i		Coranianthiis (arnoi.ni 1912)
; quot;\'S		Podostcmac. (Went 1909, 1910,
\' 5	-	1912)
	Type T):
2	Eiiphorbiac (JIodilewski 1909a,
1910, 1911, Aunom)! 1912)
Podostemac. (Magnus 1913)
to 5-	Penaeac. (Stephens 1909)
1 iquot;	Onagnic. (Geerts 1909, Moiti-
	i.ewhki iy09b, Werner 1914,
\'S	Renner 1914, Isiiikawa 1918).

Arnold!	1912	A-I-4 A-I-2a
Modilewski	•1909	AAAA-I-5
Modilewski	1910
Dessiatoff Modilewski	19H 19H	AAAA-I-5
Modilewski	19H	AAAA-I-5
Arnoldi	1912	AAAA-I-5
Treub	19H	A-IIa-3b
Rutgers	1922	A-IIa-3b
Went	1909 1910	C-I-4«
Went	1912	C-I-4«
Magnus j	1913	C-I-4
Magnus	1913	C-I-5
Magnus	1913	C-IIIa-3a
Stephens	190« 1909 1	AAAA-I-5

? ?		(1900
Campbell	1903	pathological
		(1912
? ?	Campbell	1905	pathological
5-nucleate	Campbell	1903 1912	D-IIa-3b
6-15 nucleate	Gow	1908	? ?
5-nucleate	Campbell	1912	D-IIa-3b