Textbook Brauer Groups and Obstruction Problems Moduli Spaces and Arithmetic 1St Edition Asher Auel Ebook All Chapter PDF

Brauer Groups and Obstruction
Problems Moduli Spaces and

Arithmetic 1st Edition Asher Auel
Visit to download the full and correct content document:
https://textbookfull.com/product/brauer-groups-and-obstruction-problems-moduli-spac
es-and-arithmetic-1st-edition-asher-auel/
More products digital (pdf, epub, mobi) instant
download maybe you interests ...
Algebra 1 Groups Rings Fields and Arithmetic 1st

Edition Ramji Lal
https://textbookfull.com/product/algebra-1-groups-rings-fields-
and-arithmetic-1st-edition-ramji-lal/
A Primer on Hilbert Space: Theory Linear Spaces,

Topological Spaces, Metric Spaces, Normed Spaces, and
Topological Groups 2nd Edition Carlo Alabiso
https://textbookfull.com/product/a-primer-on-hilbert-space-
theory-linear-spaces-topological-spaces-metric-spaces-normed-
spaces-and-topological-groups-2nd-edition-carlo-alabiso/
A Primer on Hilbert Space Theory Linear Spaces

Topological Spaces Metric Spaces Normed Spaces and
Topological Groups 2nd Edition Carlo Alabiso Ittay
Weiss
https://textbookfull.com/product/a-primer-on-hilbert-space-
theory-linear-spaces-topological-spaces-metric-spaces-normed-
spaces-and-topological-groups-2nd-edition-carlo-alabiso-ittay-
weiss/
A Gentle Course in Local Class Field Theory Local

Number Fields Brauer Groups Galois Cohomology 1st
Edition Pierre Guillot
https://textbookfull.com/product/a-gentle-course-in-local-class-
field-theory-local-number-fields-brauer-groups-galois-
cohomology-1st-edition-pierre-guillot/
Groups Matrices and Vector Spaces A Group Theoretic
Approach to Linear Algebra James B. Carrell
https://textbookfull.com/product/groups-matrices-and-vector-
spaces-a-group-theoretic-approach-to-linear-algebra-james-b-
carrell/
Found and Forged 2nd Edition Ivy Asher
https://textbookfull.com/product/found-and-forged-2nd-edition-
ivy-asher/
Theory of Groups and Symmetries Finite groups Lie

groups and Lie Algebras 1st Edition Alexey P. Isaev
https://textbookfull.com/product/theory-of-groups-and-symmetries-
finite-groups-lie-groups-and-lie-algebras-1st-edition-alexey-p-
isaev/
Arithmetic Lockhart
https://textbookfull.com/product/arithmetic-lockhart/
Aspects of Scattering Amplitudes and Moduli Space

Localization Sebastian Mizera
https://textbookfull.com/product/aspects-of-scattering-
amplitudes-and-moduli-space-localization-sebastian-mizera/
Progress in Mathematics
320
Asher Auel
Brendan Hassett
Anthony Várilly-Alvarado
Bianca Viray
Editors
Brauer Groups
and Obstruction
Problems
Moduli Spaces and Arithmetic
Volume 320
Series Editors
Antoine Chambert-Loir, Université Paris-Diderot, Paris, France
Jiang-Hua Lu, The University of Hong Kong, Hong Kong SAR, China
Yuri Tschinkel, Courant Institute of Mathematical Sciences, New York, USA
More information about this series at http://www.springer.com/series/4848

Asher Auel • Brendan Hassett
Anthony Várilly-Alvarado • Bianca Viray
Editors
Brauer Groups
and Obstruction Problems
Moduli Spaces and Arithmetic
Editors
Asher Auel Brendan Hassett
Department of Mathematics Department of Mathematics
Yale University Brown University
New Haven, Connecticut, USA Providence, Rhode Island, USA
Anthony Várilly-Alvarado Bianca Viray
Department of Mathematics MS-136 Department of Mathematics
Rice University University of Washington
Houston, Texas, USA Seattle, Washington, USA
ISSN 0743-1643 ISSN 2296-505X (electronic)

ISBN 978-3-319-46851-8 ISBN 978-3-319-46852-5 (eBook)
DOI 10.1007/978-3-319-46852-5
Library of Congress Control Number: 2017930680
Mathematics Subject Classification (2010): 14F05, 14F22, 14E08, 14G05, 14J28, 14J35, 14J60
© Springer International Publishing AG 2017

This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part
of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations,
recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission or
information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar
methodology now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication
does not imply, even in the absence of a specific statement, that such names are exempt from the
relevant protective laws and regulations and therefore free for general use.
The publisher, the authors and the editors are safe to assume that the advice and information in this
book are believed to be true and accurate at the date of publication. Neither the publisher nor the
authors or the editors give a warranty, express or implied, with respect to the material contained herein
or for any errors or omissions that may have been made. The publisher remains neutral with regard to
jurisdictional claims in published maps and institutional affiliations.
Printed on acid-free paper
This book is published under the trade name Birkhäuser, www.birkhauser-science.com

The registered company is Springer International Publishing AG
The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland
Contents
Nicolas Addington
The Brauer Group Is Not a Derived Invariant . . . . . . . . . . . . . . . . . . . . . . .1
Benjamin Antieau
Twisted Derived Equivalences for Affine Schemes . . . . . . . . . . . . . . . . . . . 7
Kenneth Ascher, Krishna Dasaratha, Alexander Perry, and Rong Zhou

Rational Points on Twisted K3 Surfaces and Derived Equivalences . 13
Asher Auel, Jean-Louis Colliot-Thélène, and Raman Parimala

Universal Unramified Cohomology of Cubic Fourfolds Containing
a Plane . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
Fedor Bogomolov and Yuri Tschinkel

Universal Spaces for Unramified Galois Cohomology . . . . . . . . . . . . . . . 57
Brendan Hassett and Yuri Tschinkel

Rational Points on K3 Surfaces and Derived Equivalence . . . . . . . . . . 87
Colin Ingalls, Andrew Obus, Ekin Ozman, and Bianca Viray

Unramified Brauer Classes on Cyclic Covers of the Projective
Plane . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
Martı́ Lahoz, Emanuele Macrı̀, and Paolo Stellari

Arithmetically Cohen–Macaulay Bundles on Cubic Fourfolds
Containing a Plane . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155
Kelly McKinnie, Justin Sawon, Sho Tanimoto, and

Anthony Várilly-Alvarado
Brauer Groups on K3 Surfaces and Arithmetic Applications . . . . . . 177
Alena Pirutka
On a Local-Global Principle for H 3 of Function Fields of
Surfaces over a Finite Field . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 219
Alexei N. Skorobogatov
Cohomology and the Brauer Group of Double Covers . . . . . . . . . . . . . 231
v
Preface
This volume grew out of a workshop that we organized at the American Insti-
tute of Mathematics from February 25 through March 1, 2013. The meeting
brought together experts from two different fields: number theorists inter-
ested in rational points, and complex algebraic geometers working on derived
categories of coherent sheaves. We were motivated by fresh developments
in the arithmetic of K3 surfaces, which suggest that cohomological obstruc-
tions to the existence and distribution of rational points on K3 surfaces can
be fruitfully studied via moduli spaces of twisted sheaves. Our aim was to
encourage cross-pollination between the two fields and to explore concrete
instances when the derived category of coherent sheaves on a variety over a
number field determines some of its arithmetic. Within this framework, we
aim to extend to K3 surfaces a number of powerful tools for analyzing ratio-
nal points on elliptic curves: isogenies among curves, torsion points, modular
curves, and the resulting descent techniques.
Let S1 and S2 denote complex K3 surfaces. An isomorphism

∼
ι : H 2 (S1 , Z) −→ H 2 (S2 , Z)
that is compatible with the cup product and Hodge structures causes S1 and
S2 to be isomorphic by the Torelli theorem. We might weaken this by asking
only that these lattices be stably isomorphic, or equivalently, that there is an
isomorphism between the lattices of transcendental classes
∼
T (S1 ) −→ T (S2 ). (1)
Orlov has shown that this relation coincides with the a priori algebraic notion
of derived equivalence, i.e., an equivalence of the bounded derived categories
of coherent sheaves
∼
Db (S2 ) −→ Db (S1 )
as triangulated categories over C. Such equivalences manifest themselves as
interpretations of S2 as a moduli space of sheaves over S1 and vice versa.
The manuscript by Hassett and Tschinkel explores the implications for K3
surfaces over finite and local fields. Antieau’s contribution takes a broader
perspective, contrasting the local and global features of derived equivalence
by looking at twisted derived equivalences among affine varieties.
On a K3 surface S, we arrive at the notion of cyclic isogeny by weaken-
ing (1) further to an inclusion
T (X) ,→ T (S) (2)
with cokernel isomorphic to Z/nZ, for some projective variety X. Such an
isogeny arises from, and gives rise to, an element of the Brauer group Br(S)
of S, which can be lifted to an element of the second cohomology of S with
Z/nZ coefficients. These cohomology classes on K3 surfaces are the analog
of torsion points for elliptic curves.
In 2005, van Geemen exploited the interplay between Brauer elements
and cyclic isogenies to give a complete description of the order 2 Brauer
vii
Preface viii
classes on a degree 2 K3 surface with Picard rank 1. The paper of McKinnie,

Sawon, Tanimoto, and Várilly-Alvarado generalizes van Geemen’s work to
higher order Brauer classes of higher degree K3 surfaces. Their paper uses
lattice-theoretic techniques to study the components of moduli spaces of K3
surfaces with level structure coming from the Brauer group. This is an es-
sential step toward a comprehensive geometric interpretation for the second
cohomology of a K3 surface, with applications to the structure of rational
points.
Van Geemen’s work is generalized in different directions in the contribu-
tions by Skorobogatov and by Ingalls, Obus, Ozman, and Viray. Skorobogatov
undertakes a systematic study of 2-torsion Brauer elements on double covers
of rational surfaces; Ingalls, Obus, Ozman, and Viray look at higher-degree
cyclic covers of P2 . These cases should play a central role in the study of ex-
plicit examples in the future. Both contributions handle the case of arbitrary
Picard rank.
The isogeny (2) suggests that X provides a geometric representative of
an element α ∈ Br(S)[n]. Similarly, one expects an equivalence between the
derived category of twisted sheaves Db (S, α) and an admissible subcategory
of Db (X). An important motivating example is when X is a smooth cubic
fourfold containing a plane, in which case X is birational to a quadric surface
bundle over P2 whose relative Lagrangian Grassmannian has the structure of
an étale locally trivial P1 -bundle over a K3 surface S of degree 2, giving rise to
a Brauer class α ∈ Br(S)[2]. In this case, there is an inclusion T (X) ,→ T (S),
which has index 2 when α ∈ Br(S) is nontrivial, and Kuznetsov has shown
that Db (S, α) is equivalent to an admissible subcategory of Db (X). The cubic
fourfold X is rational as soon as α ∈ Br(S)[2] becomes trivial. Using derived
category techniques, Lahoz, Macrı̀, and Stellari add to this geometric picture
by providing an elegant geometric reconstruction of the K3 surface S as a
moduli space of vector bundles on the cubic fourfold X.
Given the tight connection between Brauer groups and derived equiv-
alences for K3 surfaces, one could ask whether the Brauer group is invari-
ant under derived equivalences. Addington’s contribution answers this in the
negative by showing that the Brauer group fails to be invariant under de-
rived equivalences of Calabi–Yau threefolds. The paper of Ascher, Dasaratha,
Perry, and Zong also demonstrates the limitations of derived equivalence by
showing that naive guesses on the relation between the structure of rational
points and twisted derived equivalence for K3 surfaces are incorrect.
The key role of the Brauer group naturally leads to more systematic
analysis of higher degree unramified cohomology. Motivated by the rational-
ity problem for cubic fourfolds, Auel, Colliot-Thélène, and Parimala obtain
the universal triviality of the unramified cohomology of degree 3 for cubic
fourfolds containing a plane. More generally, inspired by a result of Merkur-
jev, they develop the notion of the universal triviality of the Chow group of
0-cycles as a possible obstruction to rationality, which has since been instru-
mental to recent breakthroughs in the (stable) rationality problem. Pirutka
ix Preface
establishes a local-global criterion for the decomposability of elements in de-

gree 3 cohomology – decompositions involving Brauer elements – for surfaces
over finite fields. Finally, Bogomolov and Tschinkel offer a broad classifying
space framework for unramified cohomology in higher degrees for varieties
over the algebraic closure of a finite field.
Acknowledgments. We are grateful to the American Institute of Mathematics
for their support of this workshop and the collaborative environment that
made this work possible. The first editor was partially supported by NSF
grant DMS-0903039 and by NSA Young Investigator grant H98230-13-1-0291.
The second editor was partially supported by NSF grant DMS-1551514. The
third editor was partially supported by NSF CAREER grant DMS-1352291.
And the fourth editor was partially supported by NSA Young Investigator
grant H98230-15-1-0054.
The Brauer Group Is Not a Derived
Invariant
Nicolas Addington
Abstract. In this short note we observe that the recent examples of

derived-equivalent Calabi–Yau 3-folds with different fundamental groups
also have different Brauer groups, using a little topological K-theory.
Mathematics Subject Classification (2010). 14F05, 14F22, 14J32.
Keywords. Brauer groups, derived equivalence, Calabi–Yau threefolds.
Some years ago Gross and Popescu [12] studied a simply-connected

Calabi–Yau 3-fold X fibered in non-principally polarized abelian surfaces.
They expected that its derived category would be equivalent to that of
the dual abelian fibration Y , which is again a Calabi–Yau 3-fold but with
π1 (Y ) = (Z8 )2 , the largest known fundamental group of any Calabi–Yau 3-
fold. This derived equivalence was later proved by Bak [2] and Schnell [23].
Ignoring the singular fibers it is just a family version of Mukai’s classic de-
rived equivalence between an abelian variety and its dual [19], but of course
the singular fibers require much more work. As Schnell pointed out, it is a
bit surprising to have derived-equivalent Calabi–Yau 3-folds with different
fundamental groups, since for example the Hodge numbers of a 3-fold are
derived invariants [22, Cor. C].
Gross and Pavanelli [11] showed that Br(X) = (Z8 )2 , the largest known
Brauer group of any Calabi–Yau 3-fold. In this note we will show that the
finite abelian group H1 (X, Z) ⊕ Br(X) is a derived invariant of Calabi–Yau
3-folds; thus in this example we must have Br(Y ) = 0, and in particular the
Brauer group alone is not a derived invariant. This too is a bit surprising, since
the Brauer group is a derived invariant of K3 surfaces: if X is a K3 surface
then Br(X) ∼ = Hom(T (X), Q/Z) [7, Lem. 5.4.1], where T (X) = N S(X)⊥ ⊂
2
H (X, Z) is the transcendental lattice, which is a derived invariant by work
of Orlov [20].
Since an earlier version of this note first circulated, Hosono and Takagi
[13] have found a second example of derived-equivalent Calabi–Yau 3-folds
with different fundamental groups. Their X and Y are constructed from
spaces of 5×5 symmetric matrices in what is likely an instance of homological
© Springer International Publishing AG 2017 1

A. Auel (eds.) et al., Brauer Groups and Obstruction Problems,
Progress in Mathematics 320, DOI 10.1007/978-3-319-46852-5_1
2 N. Addington
projective duality [14], and one has π1 (X) = Z2 and π1 (Y ) = 0. While Br(X)
and Br(Y ) are not known, from our result we see that Br(Y ) ∼ = Z2 ⊕ Br(X),
so they are different. An explicit order-2 element of Br(Y ) arises naturally in
Hosono and Takagi’s construction [13, Prop. 3.2.1].
It is worth mentioning that both π1 and Br are birational invariants, so
while birational Calabi–Yau 3-folds are derived equivalent [5], the converse
is not true. In addition to the two examples just mentioned, there is the
Pfaffian–Grassmannian derived equivalence of Borisov and Căldăraru [4]. In
that example X is a complete intersection in a Grassmannian, so H1 (X, Z) =
Br(X) = 0, so from our result we see that H1 (Y, Z) = Br(Y ) = 0 as well;
to show that X and Y are not birational Borisov and Căldăraru use a more
sophisticated minimal model program argument.
Before proving our result we fix terminology.
Definition. A Calabi–Yau 3-fold is a smooth complex projective 3-fold X
with ωX ∼ = OX and b1 (X) = 0. In particular H1 (X, Z) may be torsion.
This is in contrast to the case of surfaces, where ωX ∼
= OX and b1 (X) = 0
force π1 (X) = 0 [17, Thm. 13]. There are several reasons not to require
π1 (X) = 0 for Calabi–Yau 3-folds. As we have just seen, a simply-connected
Calabi–Yau 3-fold may be derived equivalent to a non-simply-connected one;
it may also be mirror to a non-simply-connected one. Perhaps the best reason
is that families of simply-connected and non-simply-connected Calabi–Yau 3-
folds can be connected by “extremal transitions,” that is, by performing a
birational contraction and then smoothing; most known families of Calabi–
Yau 3-folds can be connected by extremal transitions [10, 18].
Definition. The Brauer group of a smooth complex projective variety X is
2 ∗
Br(X) = tors(Han (X, OX )),
where tors denotes the torsion subgroup.
This used to be called the cohomological Brauer group until it was shown to
coincide with the honest Brauer group [8]. From the exact sequence
∗
H 2 (X, OX ) → H 2 (X, OX ) → H 3 (X, Z) → H 3 (X, OX )
we see that if X is a Calabi–Yau 3-fold then
Br(X) = tors(H 3 (X, Z)).
That is, the Brauer group of a Calabi–Yau 3-fold is entirely topological, in
contrast to that of a K3 surface which is entirely analytic.
Proposition. Let X and Y be Calabi–Yau 3-folds with Db (X) ∼ = Db (Y ). Then
∼
H1 (X, Z) ⊕ Br(X) = H1 (Y, Z) ⊕ Br(Y ).
Proof. Brunner and Distler [6, §2.5] analyzed the boundary maps in the
Atiyah–Hirzebruch spectral sequence and saw that for a Calabi–Yau 3-fold
X, or indeed any closed oriented 6-manifold with b1 (X) = 0, it degenerates
at the E2 page. Thus there is a short exact sequence
0 → H 5 (X, Z) → Ktop
1
(X) → H 3 (X, Z) → 0,
The Brauer Group Is Not a Derived Invariant 3
∗
where Ktop (X) is topological K-theory. Since H 5 (X, Z) = H1 (X, Z) is tor-
sion, this gives an exact sequence
1
0 → H1 (X, Z) → tors(Ktop (X)) → Br(X) → 0. (1)
While it is not strictly necessary for our purposes, they also got an exact
sequence
0 → Br(X)∗ → tors(Ktop 0
(X)) → H1 (X, Z)∗ → 0; (2)
∗
here if A is a finite abelian group then the dual group A := Hom(A, Q/Z),
which is non-canonically isomorphic to A.
∗
Doran and Morgan [9, §4] analyzed Ktop (X) more carefully using the
fact that c1 (X) = 0 and showed that the sequences (1) and (2) are in fact
split.
0 1
Now the proposition follows from the fact that Ktop and Ktop are de-
rived invariants [1, §2.1]. In a bit more detail, if Φ : D (X) → Db (Y ) and
b
Ψ : Db (Y ) → Db (X) are inverse equivalences, then by [20, Thm. 2.2] there

are objects E, F ∈ Db (X × Y ) such that
∗
Φ(−) = πY ∗ (E ⊗ πX (−)), Ψ(−) = πX∗ (F ⊗ πY∗ (−)),
and arguing as in [16, Lem. 5.32] we find that the same formulas define inverse
isomorphisms K ∗ (X) → K ∗ (Y ) and K ∗ (Y ) → K ∗ (X): use the fact that the
pushforward on K ∗ satisfies a projection formula and is compatible with the
pushforward on Db .
We conclude with a remark on H1 and Br in mirror symmetry. Batyrev
and Kreuzer [3] predicted that mirror symmetry exchanges H1 and Br, having
calculated both groups for all Calabi–Yau hypersurfaces in 4-dimensional
toric varieties. In all their examples the groups are quite small: either H1 = 0
and Br = Z2 , Z3 , or Z5 , or vice versa. This prediction does not seem to be
right in general. On the one hand it is contradicted by a prediction of Gross
and Pavanelli [11, Rem. 1.5], based on calculations in Pavanelli’s thesis [21],
that if X is the abelian fibration above, with H1 (X) = 0 and Br(X) = (Z8 )2 ,
then its mirror X̌ has π1 (X̌) = Br(X̌) = Z8 . Even more seriously, Hosono
and Takagi’s X and Y have the same mirror according to [15], but different
H1 and Br as we have discussed. Mirror symmetry is expected to exchange
0 1
Ktop and Ktop , however, so mirror Calabi–Yau 3-folds should have the same
H1 ⊕ Br.
Acknowledgments. I thank P. Aspinwall and A. Căldăraru for helpful dis-
cussions, and S. Hosono and H. Takagi for encouraging me to publish this
note.
References
[1] N. Addington and R. Thomas. Hodge theory and derived categories of cubic
fourfolds. Duke Math. J., 163(10):1885–1927, 2014. Also arXiv:1211.3758.
[2] A. Bak. The spectral construction for a (1,8)-polarized family of abelian vari-
eties. Preprint, arXiv:0903.5488.
4 N. Addington
[3] V. Batyrev and M. Kreuzer. Integral cohomology and mirror symmetry for
Calabi-Yau 3-folds. In Mirror symmetry. V, volume 38 of AMS/IP Stud.
Adv. Math., pages 255–270. Amer. Math. Soc., Providence, RI, 2006. Also
math/0505432.
[4] L. Borisov and A. Căldăraru. The Pfaffian-Grassmannian derived equivalence.
J. Algebraic Geom., 18(2):201–222, 2009. Also math/0608404.
[5] T. Bridgeland. Flops and derived categories. Invent. Math., 147(3):613–632,
2002. Also math/0009053.
[6] I. Brunner and J. Distler. Torsion D-branes in nongeometrical phases. Adv.
Theor. Math. Phys., 5(2):265–309, 2001. Also hep-th/0102018.
[7] A. Căldăraru. Derived categories of twisted sheaves on Calabi–Yau manifolds.
PhD thesis, Cornell University, 2000. Available at math.wisc.edu/~andreic/
publications/ThesisSingleSpaced.pdf.
[8] A. J. de Jong. A result of Gabber. Available at math.columbia.edu/~dejong/
papers/2-gabber.pdf.
[9] C. Doran and J. Morgan. Algebraic topology of Calabi-Yau threefolds in toric
varieties. Geom. Topol., 11:597–642, 2007. Also math.AG/0605074.
[10] P. Green and T. Hübsch. Connecting moduli spaces of Calabi–Yau threefolds.
Comm. Math. Phys., 119(3):431–441, 1988.
[11] M. Gross and S. Pavanelli. A Calabi-Yau threefold with Brauer group (Z/8Z)2 .
Proc. Amer. Math. Soc., 136(1):1–9, 2008. Also math/0512182.
[12] M. Gross and S. Popescu. Calabi-Yau threefolds and moduli of abelian surfaces.
I. Compositio Math., 127(2):169–228, 2001. Also math/0001089.
[13] S. Hosono and H. Takagi. Double quintic symmetroids, Reye congruences, and
their derived equivalence. Preprint, arXiv:1302.5883.
[14] S. Hosono and H. Takagi. Duality between S 2 P4 and the double quintic sym-
metroid. Preprint, arXiv:1302.5881.
[15] S. Hosono and H. Takagi. Determinantal quintics and mirror symmetry
of Reye congruences. Comm. Math. Phys., 329(3):1171–1218, 2014. Also
arXiv:1208.1813.
[16] D. Huybrechts. Fourier-Mukai transforms in algebraic geometry. Oxford Math-
ematical Monographs. The Clarendon Press Oxford University Press, Oxford,
2006.
[17] K. Kodaira. On the structure of compact complex analytic surfaces. I. Amer.
J. Math., 86:751–798, 1964.
[18] M. Kreuzer and H. Skarke. Complete classification of reflexive polyhedra in
four dimensions. Adv. Theor. Math. Phys., 4(6):1209–1230, 2000. Also hep-
th/0002240.
[19] S. Mukai. Duality between D(X) and D(X̂) with its application to Picard
sheaves. Nagoya Math. J., 81:153–175, 1981.
[20] D. Orlov. Equivalences of derived categories and K3 surfaces. J. Math. Sci.
(New York), 84(5):1361–1381, 1997. Also alg-geom/9606006.
[21] S. Pavanelli. Mirror symmetry for a two parameter family of Calabi-Yau three-
folds with Euler characteristic 0. PhD thesis, University of Warwick, 2003.
The Brauer Group Is Not a Derived Invariant 5
[22] M. Popa and C. Schnell. Derived invariance of the number of holomorphic 1-

forms and vector fields. Ann. Sci. Éc. Norm. Supér. (4), 44(3):527–536, 2011.
Also arXiv:0912.4040.
[23] C. Schnell. The fundamental group is not a derived invariant. In Derived cate-
gories in algebraic geometry, EMS Ser. Congr. Rep., pages 279–285. Eur. Math.
Soc., Zürich, 2012. Also arXiv:1112.3586.
Nicolas Addington
Department of Mathematics
University of Oregon
Eugene, OR 97403-1222
USA
e-mail: adding@uoregon.edu
Twisted Derived Equivalences for Affine
Schemes
Benjamin Antieau
Abstract. We show how work of Rickard and Toën completely resolves

the question of when two twisted affine schemes are derived equivalent.
Mathematics Subject Classification (2010). 14F22, 14F05, 16D90.
Keywords. Brauer groups and twisted derived equivalences.
1. Introduction
The question of when Db (X) is equivalent as a k-linear triangulated category
to Db (Y ) for two varieties X and Y over a field k has been extensively studied
since Mukai proved that Db (Â) ' Db (A) for an abelian variety A and its dual
Â [8]. Since in general A and Â are not isomorphic, derived equivalence is a
weaker condition than isomorphism. However, derived equivalence nevertheless
does preserve a great deal of information: derived equivalent varieties have
the same dimension, the same algebraic K-theory, and the same Hochschild
homology.
The cohomological Brauer group of a scheme X is
Br0 (X) = Hét
2
(X, Gm )tors .
When X is quasi-compact, there is an inclusion Br(X) ⊆ Br0 (X), where
Br(X) denotes the Brauer group of X, which classifies Brauer equivalence
classes of Azumaya algebras on X. In many cases of interest, Br(X) = Br0 (X).
Examples include all quasi-projective schemes over affine schemes [3]. The
Brauer group comes into play because in many problems on moduli of vector
bundles, there is an obstruction, living in the Brauer group of the coarse
moduli space, to the existence of a universal vector bundle. Another way to
say this is that this class in the Brauer group is the obstruction to the coarse
moduli space being fine. See [1, Introduction]. At times one then obtains
an equivalence Db (X) ' Db (Y, β), where Db (Y, β) is the derived category of
β-twisted coherent sheaves. Particular cases of this arise in the study of K3

8 B. Antieau
surfaces for example, and especially in the work of Huybrechts, Macrı̀, and
Stellari [4–6].
The systematic study of when Db (X, α) ' Db (Y, β) began with Căldă-
raru’s thesis [1]. In this short note, we are interested in the following two
problems.
Problem 1.1. Let R be a commutative ring, let X and Y be two locally
noetherian R-schemes, and fix α ∈ Br0 (X) and β ∈ Br0 (Y ). Determine when
there exists an R-linear equivalence of triangulated categories Db (X, α) '
Db (Y, β).
Problem 1.2. Let R be a commutative ring, let X and Y be two quasi-
compact and quasi-separated R-schemes, and fix α ∈ Br0 (X) and β ∈ Br0 (Y ).
Determine when there exists an R-linear equivalence of triangulated categories
Dperf (X, α) ' Dperf (Y, β).
Here, Db (X, α) denotes the bounded derived category of α-twisted coher-
ent sheaves, while Dperf (X, α) is the triangulated category of perfect complexes
of α-twisted OX -modules. When X is regular and noetherian, the existence lo-
cally of finite-length finitely generated locally free resolutions implies that the
natural map Dperf (X, α) → Db (X, α) is an equivalence of R-linear triangulated
categories.
Question 1.3. Are Problems 1.1 and 1.2 equivalent for X and Y locally
noetherian, quasi-compact, and quasi-separated?
The contents of our paper are as follows. In Section 2, we give some
background on twisted derived categories and equivalences between them.
Then, in Section 3, the affine case of Problems 1.1 and 1.2 is solved completely,
and we explain how work of Rickard shows that these two problems are
equivalent for affine schemes. We do not claim that this result is new, but
rather that it is not as well-known as it should be.
Acknowledgments. We would like to thank the organizers of the AIM work-
shop “Brauer groups and obstruction problems” for facilitating a stimulating
week in February 2013.
2. Twisted derived categories

Let X be a scheme, and take α ∈ Hét 2
(X, Gm ). Then, α is represented by
a Gm -gerbe X → X. There is a good notion of quasi-coherent sheaf on X,
or of coherent sheaf when X is locally noetherian. An OX -module F comes
naturally with a left action of the sheaf Gm,X . Moreover, there is a second,
inertial action, which can be described as saying that a section u ∈ Gm,X (U )
over U → X acts on F(U ) via the isomorphism u∗ FU → FU , which induces
an isomorphism u∗ : F(U ) → F(U ). There is an associated left action of the
inertial action. An α-twisted OX -module is by definition an OX -module F for
which these two left actions agree. It is shown in [7, Proposition 2.1.3.3] that
this agrees with the definition of α-twisted sheaf given by Căldăraru.
Twisted Derived Equivalences for Affine Schemes 9
If X is a scheme and α ∈ Hét 2

(X, Gm ), write Dperf (X, α) for the tri-
angulated category of complexes of α-twisted sheaves that are étale locally
quasi-isomorphic to finite-length complexes of vector bundles. This is natu-
rally a full subcategory of Dqc (X, α) of complexes of α-twisted sheaves with
(α-twisted) quasi-coherent cohomology sheaves. If X is regular and noetherian,
then Dperf (X, α) ' Db (X, α), the bounded derived category of α-twisted
coherent OX -modules.
Let A be an Azumaya algebra on X with class α (so that α ∈ Br(X)
in this case). A complex of right A-modules P (in the abelian category
ModOX ) is perfect if there is an open affine cover {Ui }i∈I of X such that
PUi is quasi-isomorphic to a bounded complex of finitely generated projective
right Γ(Ui , A)-modules. The derived category of perfect complexes of right A-
modules will be denoted Dperf (X, A). Then, as explained in [1], Dperf (X, α) '
Dperf (X, A). In the same way, there is a big derived category of all complexes
of right A-modules with quasi-coherent cohomology sheaves Dqc (X, A) and
an equivalence Dqc (X, α) ' Dqc (X, A).
In the next section, we will need dg enhancements of these categories.
Write Perf(X, α) and QC(X, α) for dg enhancements of Dperf (X, α) and
Dqc (X, α), respectively. These are pretriangulated dg categories. The big dg
category QC(X, α) is constructed for example in Toën [12]. The small dg
category Perf(X, α) can then be taken to be the dg category of compact
objects in QC(X, α).
3. Twisted derived equivalences over affine schemes

Many of us first learned of twisted derived categories from Căldăraru’s thesis [1]
and the paper [2]. In that paper, Căldăraru cites a private communication
from Yekutieli giving the following theorem [1, Theorem 6.2].
Theorem 3.1. Suppose that R is a commutative local ring and that A and B
are Azumaya R-algebras with classes α, β ∈ Br(R). Then, the following are
equivalent:
1. α = β in Br(R);
2. A and B are derived Morita equivalent over R – that is, there is an
R-linear equivalence of triangulated categories Db (A) ' Db (B).
It is the purpose of our paper to advertise the fact that the condition
that R be local is unnecessary.
Theorem 3.2. Suppose that R is a commutative ring and that α, β ∈ Br(R).
Then, the following are equivalent:
1. α = β in Br(R);
2. there is an R-linear equivalence of triangulated categories Dperf (R, α) '
Dperf (R, β).
Moreover, if R is noetherian, these are equivalent to:
10 B. Antieau
3. there is an R-linear equivalence of triangulated categories Db (R, α) '

Db (R, β).
Proof. Let X = Spec R. Since α, β ∈ Br(R), we can assume that α is repre-
sented by an Azumaya R-algebra A, and that β is represented by an Azumaya
R-algebra B. In this case, Dperf (X, α) ' Dperf (X, A) ' Db (projA ), where
Db (projA ) is the bounded derived category of finitely generated projective
A-modules. The second equivalence follows because on an affine scheme, every
perfect complex is quasi-isomorphic to a bounded complex of finitely gen-
erated projectives (see [11, Proposition 2.3.1(d)]), and this generalizes in a
straightforward way to Azumaya algebras on an affine scheme.
When α = β, the Azumaya algebras A and B are Brauer equivalent.
This means that there exist finitely generated projective R-modules M and
N and an R-algebra isomorphism
A ⊗R EndR (M ) ∼ = B ⊗R EndR (N ).
It follows from classical Morita theory that there is an equivalence ModA '
ModB of abelian categories of right A and B-modules. From this it follows
immediately that Db (projA ) ' Db (projB ). This proves that (1) implies (2).
So, suppose that Dperf (R, α) ' Dperf (R, β), or in other words that
Db (projA ) ' Db (projB ). Rickard’s theorem [9, Theorem 6.4] as refined in [10]
implies that there is a tilting complex inducing an R-linear equivalence
Db (projA ) ' Db (projB ). (Rickard’s theorem does not imply that this is the
equivalence we began with, but it is still R-linear.) The existence of the
tilting complex implies that there is an equivalence of R-linear dg category
enhancements Perf(R, α) ' Perf(R, β), which is then a derived Morita equiv-
alence. Recall that QC(R, α) is equivalent to the derived category of dg
modules over Perf(R, α), and similarly for QC(R, β). Hence, the equivalence
of small dg categories Perf(R, α) ' Perf(R, β) induces an equivalence of
the “big” R-linear dg categories QC(R, α) ' QC(R, β). These are locally
presentable dg categories with descent in the language of [12]. Now, the
derived Brauer group of R, denoted dBr(R), classifies locally presentable dg
categories with descent over R that are étale locally equivalent to QC(R).
Since Spec R is affine, the R-linear equivalence QC(R, α) ' QC(R, β) means
that α and β define the same element of dBr(R) (see [12, Section 3]). But,
dBr(R) ∼ = Hét1
(Spec R, Z) × Hét
2
(Spec R, Gm ) by [12, Theorem 1.1]. Since
Br(R) ⊆ dBr(R), it follows that α = β, and so (2) implies (1).
Finally, the fact that (2) and (3) are equivalent follows from [9, Proposi-
tions 8.1, 8.2]. This completes the proof.
Remark 3.3. By [3], Br(R) = Br0 (R) = Hét

2
(Spec R, Gm )tors .
We expand briefly on the philosophy of the proof. Write QC for the étale
stack of locally presentable dg categories with dg category of sections over
f : Y → X the locally presentable dg category QC(Y ), which is a dg category
enhancement of the derived category Dqc (Y ) of complexes of OY -modules
with quasi-coherent cohomology sheaves. The derived Brauer group dBr(X)
Twisted Derived Equivalences for Affine Schemes 11
of a scheme classifies stacks of locally presentable dg categories that are étale

locally equivalent to QC up to equivalence of stacks.
Motto 3.4. The Brauer group classifies Azumaya algebras A up to derived
Morita equivalence of stacks of dg categories of complexes of A-modules.
For α ∈ dBr(X), write QC(α) for the associated stack. For instance, if
α is the Brauer class of an Azumaya algebra A over X then the dg category
of sections over f : Y → X of QC(α) is QC(Y, f ∗ A), which is a dg category
enhancement of Dqc (Y, A) ' Dqc (Y, α). The key point in the proof of the
theorem was that over an affine scheme Spec R, giving an equivalence of stacks
QC(α) ' QC(β) is equivalent to giving an R-linear equivalence of the global
sections QC(R, α) ' QC(R, β).
On non-affine schemes, giving an equivalence of global sections is, not
surprisingly, insufficient. The following example is due to Căldăraru [1, Ex-
ample 1.3.16]. Let X be a smooth projective K3 surface over the complex
numbers given as a double cover of P2 branched along a smooth sextic curve.
The involution φ of X given by interchanging the sheets of the cover has the
property that φ∗ α = −α for α ∈ Br(X). Clearly φ induces an equivalence
Db (X, α) ' Db (X, −α). But, since Br(X) contains non-zero p-torsion for every
prime p, there is a class α ∈ Br(X) such that α 6= −α. Thus, the theorem fails
in the non-affine case. The problem is that the equivalence does not respect
restriction to open subsets of X.
We now prove the conjecture suggested by Căldăraru after [2, Theorem
6.2].
Corollary 3.5. Suppose that R and S are commutative rings and that there
is an equivalence of triangulated categories Dperf (R, α) ' Dperf (S, β) for
α ∈ Br(R) and β ∈ Br(S). Then, there exists a ring isomorphism φ : R → S
such that φ∗ (α) = β in Br(S).
Proof. Let A be an Azumaya algebra with class α over R, and let B be
an Azumaya algebra over S with class β. Then, our hypotheses say that
Db (projA ) ' Db (projB ). By Rickard [9, Proposition 9.2], the centers of A
and B are isomorphic. As before, by Rickard [9, Proposition 9.2], there
is an isomorphism φ : R → S, and there are equivalences Dperf (S, β) '
Dperf (R, α) ' Dperf (S, φ∗ (α)). The composition induces a ring automorphism
σ : S → S. So, by composing on the φ on the right with σ −1 , we can assume
that Dperf (S, β) ' Dperf (S, φ∗ (α)) is S-linear. The corollary follows now from
Theorem 3.2.
We end by observing that the condition of R-linearity in Theorem 3.2 is

necessary.
Remark 3.6. Consider the field k = C(w, x, y, z) and the quaternion divi-
sion algebras (w, x) and (y, z) over k. Then, these algebras are evidently
derived Morita equivalent over C (they are even isomorphic over C). However,
[(w, x)] 6= [(y, z)] in Br(k).
12 B. Antieau
References
[1] A. Căldăraru, Derived categories of twisted sheaves on Calabi-Yau manifolds. Ph.D.
thesis, Cornell University (2000), available at http://www.math.wisc.edu/~andreic/.
[2] , Derived categories of twisted sheaves on elliptic threefolds, J. Reine Angew.
Math. 544 (2002), 161–179.
[3] A. J. de Jong, A result of Gabber, available at http://www.math.columbia.edu/
~dejong/.
[4] D. Huybrechts and P. Stellari, Equivalences of twisted K3 surfaces, Math. Ann. 332
(2005), no. 4, 901–936.
[5] , Proof of Căldăraru’s conjecture. Appendix: “Moduli spaces of twisted sheaves
on a projective variety” by K. Yoshioka, Moduli spaces and arithmetic geometry, Adv.
Stud. Pure Math., vol. 45, Math. Soc. Japan, Tokyo, 2006, pp. 31–42.
[6] D. Huybrechts, E. Macrı̀, and P. Stellari, Derived equivalences of K3 surfaces and
orientation, Duke Math. J. 149 (2009), no. 3, 461–507.
[7] M. Lieblich, Moduli of twisted sheaves, Duke Math. J. 138 (2007), no. 1, 23–118.
[8] S. Mukai, Duality between D(X) and D(X̂) with its application to Picard sheaves,
Nagoya Math. J. 81 (1981), 153–175.
[9] J. Rickard, Morita theory for derived categories, J. London Math. Soc. (2) 39 (1989),
no. 3, 436–456.
[10] , Derived equivalences as derived functors, J. London Math. Soc. (2) 43 (1991),
no. 1, 37–48.
[11] R. W. Thomason and T. Trobaugh, Higher algebraic K-theory of schemes and of de-
rived categories, The Grothendieck Festschrift, Vol. III, Progr. Math., vol. 88, Birkhäuser
Boston, Boston, MA, 1990, pp. 247–435.
[12] B. Toën, Derived Azumaya algebras and generators for twisted derived categories,
Invent. Math. 189 (2012), no. 3, 581–652.
Benjamin Antieau
Department of Mathematics, Statistics, and Computer Science
University of Illinois at Chicago
851 S. Morgan St.
Chicago, IL 60607
USA
e-mail: benjamin.antieau@gmail.com
Rational Points on Twisted K3 Surfaces
and Derived Equivalences
Kenneth Ascher, Krishna Dasaratha, Alexander Perry and
Rong Zhou
Abstract. Using a construction of Hassett and Várilly-Alvarado, we pro-

duce derived equivalent twisted K3 surfaces over Q, Q2 , and R, where
one has a rational point and the other does not. This answers negatively
a question recently raised by Hassett and Tschinkel.
Mathematics Subject Classification (2010). 14G05, 14J28, 18E30.
Keywords. Rational points, twisted K3 surfaces, derived categories.
1. Introduction
A twisted K3 surface is a pair (X, α), where X is a K3 surface and α ∈ Br(X)
is a Brauer class. In a recent survey paper [5], Hassett and Tschinkel asked
whether the existence of a rational point on a twisted K3 surface is invariant
under derived equivalence. More precisely, they asked:
Question. Let (X1 , α1 ) and (X2 , α2 ) be twisted K3 surfaces over a field k.
Suppose there is a k-linear equivalence
Db (X1 , α1 ) ' Db (X2 , α2 )
of twisted derived categories. Then is the existence of a k-point of (X1 , α1 )
equivalent to the existence of a k-point of (X2 , α2 )?
By definition, a k-point of a twisted K3 surface (X, α) is a point x ∈ X(k)
such that the evaluation α(x) = 0 ∈ Br(k). Equivalently, it is a k-point of
the Gm -gerbe over X associated to α.
In [5], it is shown that for the untwisted case of the question where
α1 , α2 vanish, the answer is positive over certain fields k, e.g., R, finite fields,
and p-adic fields (provided the Xi have good reduction, or p ≥ 7 and the Xi
This project was supported by the 2015 AWS and NSF grant DMS-1161523. K.A. was
also partially supported by NSF grant DMS-1162367. A.P. was also partially supported by
NSF GRFP grant DGE1144152.

14 K. Ascher, K. Dasaratha, A. Perry and R. Zhou
have ADE reduction). The purpose of this paper is to show that if α1 , α2 are
allowed to be nontrivial, the answer to the question is negative for k = Q, Q2 ,
and R.
We work over a field k of characteristic not equal to 2, and consider a
double cover Y → P2 × P2 ramified over a divisor of bidegree (2, 2). The
projection πi : Y → P2 onto the i-th P2 factor, i = 1, 2, realizes Y as a
quadric fibration. Provided that the discriminant divisor of πi is smooth, the
Stein factorization of the relative Fano variety of lines of πi is a K3 surface
Xi , which comes with a natural Brauer class αi . In this setup, we prove the
following result.
Theorem 1.1. There is a k-linear equivalence Db (X1 , α1 ) ' Db (X2 , α2 ).
We note that this result seems to be known to the experts (at least for
k = C), but we could not find a proof in the literature.
Hassett and Várilly-Alvarado studied the above construction of twisted
K3s in relation to rational points [6]. They show that over k = Q, if certain
conditions are imposed on the branch divisor Z ⊂ P2 × P2 of Y , the class
α1 gives a (transcendental) Brauer–Manin obstruction to the Hasse principle
on X1 . A priori, α2 need not obstruct the existence of rational points on X2 .
In fact, it is possible that X2 has rational points, but the conditions imposed
on Z result in very large coefficients of the defining equation of X2 , making
a computer search for points infeasible.
In this paper, we observe that the 2-adic condition imposed by Has-
sett and Várilly-Alvarado can be relaxed, while still guaranteeing α1 gives a
Brauer–Manin obstruction (see Lemma 4.5). The upshot is that the defining
coefficients of X2 are much smaller, making it easy to find rational points
with a computer. Up to modifying the αi by a Brauer class pulled back from
k = Q, we obtain the desired example over Q. We also check the example
“localizes” over Q2 and R. More precisely, we prove:
Theorem 1.2. For k = Q, Q2 , or R, the divisor Z ⊂ P2 × P2 can be cho-
sen so that there are Brauer classes αi0 ∈ Br(Xi ), congruent to αi modulo
Im(Br(k) → Br(Xi )), such that:
1. There is a k-linear equivalence Db (X1 , α10 ) ' Db (X2 , α20 ),
2. (X1 , α10 ) has no k-point,
3. (X2 , α20 ) has a k-point.
Acknowledgements. The authors thank the organizers of the 2015 Arizona
Winter School, where this work began. We thank Asher Auel, Brendan Has-
sett, and Sho Tanimoto for useful discussions. Above all, we thank Tony
Várilly-Alvarado for suggesting this project, and for his help and encourage-
ment along the way.
2. Construction of the twisted K3 surfaces

In this section, k denotes a base field of characteristic not equal to 2.
Rational Points on Twisted K3 Surfaces and Derived Equivalences 15
2.1. Quadric fibrations

We start by reviewing some terminology on quadric fibrations. Let S be a
variety over k, i.e., an integral, separated scheme of finite type over k. Let E
be a rank n ≥ 2 vector bundle on S, i.e., a locally free OS -module of rank n.
Our convention is that the projective bundle of E is the morphism
p : P(E) = ProjS (Sym• (E∗ )) → S.
A quadric fibration is determined by a line bundle L on S and a nonzero
section
s ∈ Γ(P(E), OP(E) (2) ⊗ p∗ L) = Γ(S, Sym2 (E∗ ) ⊗ L).
Namely, the zero locus of s on P(E) defines a subvariety Q, and the restriction
π : Q → S of p : P(E) → S is the associated quadric fibration, which if flat
is of relative dimension n − 2. Below we will be specifically interested in flat
quadric fibrations of relative dimension 2, which we refer to as quadric surface
fibrations.
Note that the section of Sym2 (E∗ ) ⊗ L defining a quadric fibration cor-
responds to a morphism q : E → E∗ ⊗ L. Taking the determinant gives rise
to a section of det(E∗ )2 ⊗ Ln whose vanishing defines the discriminant locus
D ⊂ S, which is a divisor provided π : Q → S is generically smooth. The
fibration π : Q → S is said to have simple degeneration if the fiber over every
closed point of S is a quadric of corank ≤ 1. We note that if π : Q → S is
flat and generically smooth and S is smooth over k, then the discriminant
divisor D is smooth over k if and only if Q is smooth over k and π has simple
degeneration [1, Proposition 1.6].
2.2. Twisted K3 surfaces

Let V1 and V2 be 3-dimensional vector spaces over k. We denote by Hi the
hyperplane class on P(Vi ); by abuse of notation, we denote by the same letter
the pullback of Hi to any variety mapping to P(Vi ). Let
π : Y → P(V1 ) × P(V2 )
be the double cover of P(V1 ) × P(V2 ) ramified over a smooth divisor Z in
the linear system |2H1 + 2H2 |. Let pri : P(V1 ) × P(V2 ) → P(Vi ) be the i-th
projection, and let πi = pri ◦ π : Y → P(Vi ).
Lemma 2.1. Let E1 = (V2 ⊗ O) ⊕ O(H1 ) on P(V1 ) and E2 = (V1 ⊗ O) ⊕ O(H2 )
on P(V2 ). Then for i = 1, 2 there is a commutative diagram
Y
ji
/ P(Ei )
πi
pi
{
P(Vi )
where ji is a closed immersion, j1∗ OP(E1 ) (1) = OY (H2 ), j2∗ OP(E2 ) (1) = OY (H1 ).
Moreover, Y is cut out on P(Ei ) by a section of OP(Ei ) (2) ⊗ O(2Hi ), so that
πi is a quadric surface fibration.
Proof. Consider the case i = 1. The morphism j1 : Y → P(E1 ) is given

by the π1 -very ample line bundle OY (H2 ). More precisely, using that
π∗ (OY ) = O ⊕ O(−H1 − H2 ), we find
π1∗ (OY (H2 )) = pr1∗ (O(H2 ) ⊕ O(−H1 ))
= (V2∗ ⊗ O) ⊕ O(−H1 )
= E∗1 .
Working locally on P(V1 ), we see the canonical map
π1∗ E∗1 = π1∗ π1∗ (OY (H2 )) → OY (H2 )
is surjective and the corresponding morphism j1 : Y → P(E1 ) is an immer-
sion. By construction j1∗ OP(E1 ) (1) = OY (H2 ). Moreover, if ζ denotes the class
of OP(E1 ) (1) in Pic(P(E1 )), then it is easy to compute
[Y ] = 2ζ + 2H1 ∈ Pic(P(E1 ))
by using the intersection numbers H12 H22 = 2 and H1 H23 = 0 on Y . So Y is
indeed a quadric surface fibration, cut out by a section of OP(E1 ) (2) ⊗ O(2H1 )
on P(E1 ).
Let Di denote the discriminant divisor of πi : Y → P(Vi ). It follows from
the lemma that Di is defined by a section of det(E∗i )2 ⊗ O(8Hi ) = O(6Hi ),
i.e., Di ⊂ P(Vi ) is a sextic curve. Let fi : Xi → P(Vi ) be the double cover
of P(Vi ) ramified over Di . If Di is smooth (equivalently, if πi has simple
degeneration), then Xi is a smooth K3 surface. Moreover, Xi comes equipped
with an Azumaya algebra Ai , as follows.
In general, suppose given a generically smooth quadric surface fibration
π : Q → S over a smooth k-variety S, with smooth discriminant divisor and
simple degeneration. Let F → S be the relative Fano variety of lines of π. It
follows from [7, Proposition 3.3] that Stein factorization gives morphisms
g f
F −→ X −→ S,
where g is an étale locally trivial P1 -bundle over X and f is the double cover
of S branched along the discriminant divisor D. The morphism g corresponds
to an Azumaya algebra A on X.
Applying this discussion to πi : Y → P(Vi ), we see that if Di is smooth,
then Xi is equipped with an Azumaya algebra Ai . Of course Ai represents a
Brauer class αi ∈ Br(Xi ), so we can regard the pair (Xi , Ai ) as a twisted K3
surface.
3. Derived equivalence of the twisted K3 surfaces

In this section, we prove the twisted K3 surfaces (Xi , Ai ) of the previous sec-
tion are derived equivalent. Our proof works over any field k of characteristic
not equal to 2, and gives an explicit functor inducing the equivalence. The
key tool is Kuznetsov’s semiorthogonal decomposition of the derived category
of a quadric fibration [9].
3.1. Conventions
All triangulated categories appearing below will be k-linear, and functors
between them will be k-linear and exact.
For a variety X, we denote by Db (X) the bounded derived category of
coherent sheaves on X, regarded as a triangulated category. More generally,
for any sheaf of OX -algebras A which is coherent as an OX -module, we denote
by Db (X, A) the bounded derived category of coherent sheaves of right A-
modules on X. We note that if A is an Azumaya algebra corresponding to
a Brauer class α ∈ Br(X), then the bounded derived category of α-twisted
sheaves Db (X, α) is equivalent to Db (X, A).
As a rule, all functors we consider are derived. More precisely, for a
morphism of varieties f : X → Y , we simply write f∗ : Db (X) → Db (Y ) for
the derived pushforward (provided f is proper) and f ∗ : Db (Y ) → Db (X)
for the derived pullback (provided f has finite Tor-dimension). Similarly, for
F, G ∈ Db (X), we write F ⊗ G ∈ Db (X) for the derived tensor product.
3.2. Semiorthogonal decompositions
One way to understand the derived category of a variety (or more generally
a triangulated category) is by “decomposing” it into simpler pieces. This is
formalized by the notion of a semiorthogonal decomposition, which plays a
central role in the rest of this section. We summarize the rudiments of this
theory; see, e.g., [3] and [4] for a more detailed exposition.
Definition 3.1. Let T be a triangulated category. A semiorthogonal decompo-
sition
T = hA1 , . . . , An i
is a sequence of full triangulated subcategories A1 , . . . , An of T – called the
components of the decomposition – such that:
1. Hom(F, G) = 0 for all F ∈ Ai , G ∈ Aj if i > j.
2. For any F ∈ T, there is a sequence of morphisms
0 = Fn → Fn−1 → · · · → F1 → F0 = F,
such that Cone(Fi → Fi−1 ) ∈ Ai .
Semiorthogonal decompositions are closely related to the notion of an
admissible subcategory of a triangulated category. Such a subcategory A ⊂ T
is by definition a full triangulated subcategory whose inclusion i : A ,→ T
admits right and left adjoints i! : T → A and i∗ : T → A. For X a smooth
proper variety over k, the components of any semiorthogonal decomposition
of Db (X) are in fact admissible subcategories.
The simplest examples of admissible subcategories come from excep-
tional objects. An object F ∈ T of a triangulated category is called exceptional
if
k if p = 0,
Hom(F, F[p]) =
0 if p 6= 0.
If X is a proper variety and F ∈ Db (X) is exceptional, then the full triangu-
lated subcategory hFi ⊂ Db (X) generated by F is admissible and equivalent
to the derived category of a point via Db (Spec(k)) → Db (X) : V 7→ V ⊗F. To

simplify notation, we write F in place of hFi when hFi appears as a component
in a semiorthogonal decomposition, i.e., instead of Db (X) = h. . . , hFi, . . . i we
write Db (X) = h. . . , F, . . . i.
Example 3.2. It is easy to see any line bundle on projective space Pn is
exceptional as an object of Db (Pn ). In fact, Beilinson [2] showed Db (Pn ) has
a semiorthogonal decomposition into n + 1 line bundles, namely
Db (Pn ) = hO, O(1), . . . , O(n)i.
Given one semiorthogonal decomposition of a triangulated category T,
others can be obtained via mutation functors. If i : A ,→ T is the inclusion of
an admissible subcategory, the left and right mutation functors LA : T → T
and RA : T → T are defined by the formulas
LA (F) = Cone(ii! F → F) and RA (F) = Cone(F → ii∗ F)[−1],
where ii! F → F and F → ii∗ F are the counit and unit morphisms of the
adjunctions. These functors satisfy the following basic properties.
Lemma 3.3. The mutation functors LA and RA annihilate A. Moreover, they
restrict to mutually inverse equivalences
∼ ∼
LA |⊥ A : ⊥ A −
→ A⊥ and RA |A⊥ : A⊥ −
→ ⊥ A,
where A⊥ and ⊥ A are the right and left orthogonal categories to A, i.e., the
full subcategories of T defined by
A⊥ = {F ∈ T | Hom(G, F) = 0 for all G ∈ A} ,
⊥
A = {F ∈ T | Hom(F, G) = 0 for all G ∈ A} .
The following lemma describes the action of mutation functors on a
semiorthogonal decomposition.
Lemma 3.4. Let T = hA1 , . . . , An i be a semiorthogonal decomposition with
admissible components. Then for 1 ≤ i ≤ n − 1 there is a semiorthogonal
decomposition
T = hA1 , . . . , Ai−1 , LAi (Ai+1 ), Ai , Ai+2 , . . . , An i,
and for 2 ≤ i ≤ n there is a semiorthogonal decomposition
T = hA1 , . . . , Ai−2 , Ai , RAi (Ai−1 ), Ai+1 , . . . , An i.
We will also need the following lemma, which allows us to compute the
effect of a mutation functor in a special case. It follows easily from Serre
duality.
Lemma 3.5. Let X be a smooth projective variety over k. Assume that we
are given a semiorthogonal decomposition Db (X) = hA1 , . . . , An i. Then we
have LhA1 ,...,An−1 i (An ) = An ⊗ ωX , where An ⊗ ωX denotes the image of An
under the autoequivalence F 7→ F ⊗ ωX of Db (X).
3.3. Derived categories of quadric fibrations

Let π : Q → S be a quadric fibration associated to a rank n vector bundle E
and a section of Sym2 (E∗ ) ⊗ L, as in Section 2.1. Then there is an associated
even Clifford algebra C`0 , which is a sheaf of algebras on S given as a certain
quotient of the tensor algebra T• (E ⊗ E ⊗ L∗ ). For the precise definition,
see [1, Section 1.5] (cf. [9, Section 3.3]). We note that C`0 admits an OS -
module filtration of length b n2 c with associated graded pieces ∧2i E ⊗ (L∗ )i .
In case the fibration π : Q → S is flat and S is smooth over k,
Kuznestov [9] established a semiorthogonal decomposition of Db (Q) into a
copy of Db (S, C`0 ) and a number of copies of Db (S). In fact, Kuznetsov stated
his result under the assumption that k is algebraically closed of character-
istic 0, but as explained in [1, Theorem 2.11], the proof works without this
hypothesis.
Theorem 3.6 ([9, Theorem 4.2]). Let π : Q → S be a flat quadric fibration
of relative dimension n − 2 over a smooth k-variety S. Let OQ (1) denote the
restriction of OP(E) (1) to Q. Then the functor π ∗ : Db (S) → Db (Q) is fully
faithful, and there is a fully faithful functor Φ : Db (S, C`0 ) → Db (Q) such
that there is a semiorthogonal decomposition
Db (Q) = hΦ(Db (S, C`0 )), π ∗ Db (S) ⊗ OQ (1), . . . , π ∗ Db (S) ⊗ OQ (n − 2)i.
Remark 3.7. The functor Φ : Db (S, C`0 ) → Db (Q) is given by an explicit
Fourier–Mukai kernel, see [9, Section 4].
Now assume π : Q → S is a generically smooth quadric surface fibra-
tion over a smooth k-variety S, with smooth discriminant divisor and simple
degeneration. As in the discussion at the end of Section 2.2, the double cover
f : X → S ramified over D is equipped with an Azumaya algebra A. In terms
of this data, we have the following alternative description of Db (S, C`0 ), see [1,
Proposition B.3] or [10, Lemma 4.2].
Lemma 3.8. In the above situation, there is an isomorphism f∗ A ∼ = C`0 . In
∼
particular, there is an equivalence f∗ : Db (X, A) −
→ Db (S, C`0 ).
3.4. Derived equivalence
Let π : Y → P(V1 ) × P(V2 ) be as in Section 2.2. Assume the discriminant
divisors Di of the quadric fibrations πi : Y → P(Vi ) are smooth, so that
we get associated twisted K3 surfaces (Xi , Ai ). Let C`0,i denote the even
Clifford algebra of the quadric fibration πi : Y → P(Vi ). Then Lemma 3.8
∼
gives an equivalence fi∗ : Db (Xi , Ai ) −
→ Db (P(Vi ), C`0,i ). Finally, denote by
b b
Φi : D (P(Vi ), C`0,i ) → D (Y ) the fully faithful functor from Theorem 3.6.
In this setup, we prove the following result.
Theorem 3.9. Assume D1 and D2 are smooth. Then there is an equivalence
Db (X1 , A1 ) ' Db (X2 , A2 )
given by the composition
−1
f2∗ ◦ Φ∗2 ◦ ROY (H2 ) ◦ LOY (H1 ) ◦ Φ1 ◦ f1∗ : Db (X1 , A1 ) → Db (X2 , A2 ),
where
• LOY (H1 ) is the left mutation functor through hOY (H1 )i ⊂ Db (Y ),
• ROY (H2 ) is the right mutation functor through hOY (H2 )i ⊂ Db (Y ),
• Φ∗2 is the left adjoint of Φ2 ,
−1
• f2∗ is the inverse of the equivalence
∼
f2∗ : Db (X2 , A2 ) −
→ Db (P(V2 ), C`0,2 ).
The theorem is an immediate consequence of the following proposition.
We note that the proposition holds without assuming smoothness of the
discriminant divisors Di .
Proposition 3.10. There is an equivalence
Db (P(V1 ), C`0,1 ) ' Db (P(V2 ), C`0,2 )
given by the composition
Φ∗2 ◦ ROY (H2 ) ◦ LOY (H1 ) ◦ Φ1 : Db (P(V1 ), C`0,1 ) → Db (P(V2 ), C`0,2 ).
Proof. Set Ci = Φi (Db (P(Vi ), C`0,i )) ⊂ Db (Y ). Theorem 3.6 gives semior-

thogonal decompositions
Db (Y ) = hC1 , π1∗ Db (P(V1 )) ⊗ O(H2 ), π1∗ Db (P(V1 )) ⊗ O(2H2 )i,
Db (Y ) = hC2 , π2∗ Db (P(V2 )) ⊗ O(H1 ), π2∗ Db (P(V2 )) ⊗ O(2H1 )i.
Recall Beilinson’s decomposition Db (P(Vi )) = hO, O(Hi ), O(2Hi )i (see Ex-

ample 3.2). In each of the above decompositions of Db (Y ), we replace the
first copy of Db (P(Vi )) by Beilinson’s decomposition and the second copy by
the same decomposition twisted by O(Hi ):
Db (Y ) = hC1 , O(H2 ), O(H1 + H2 ), O(2H1 + H2 ),
(3.1)
O(H1 + 2H2 ), O(2H1 + 2H2 ), O(3H1 + 2H2 )i,
Db (Y ) = hC2 , O(H1 ), O(H1 + H2 ), O(H1 + 2H2 ),
(3.2)
O(2H1 + H2 ), O(2H1 + 2H2 ), O(2H1 + 3H2 )i.
We perform a sequence of mutations that identifies the categories generated
by the exceptional objects in (3.1) and (3.2).
First consider (3.1). Mutate O(3H1 + 2H2 ) to the far left of the decom-
position. Note that Y is smooth with canonical class KY = −2H1 − 2H2 , so
by Lemma 3.5 the result of the mutation is
Db (Y ) = hO(H1 ), C1 , O(H2 ), O(H1 + H2 ), O(2H1 + H2 ),
O(H1 + 2H2 ), O(2H1 + 2H2 )i.
Left mutating C1 through O(H1 ) then gives a decomposition
Db (Y ) = hLO(H1 ) C1 , O(H1 ), O(H2 ), O(H1 + H2 ), O(2H1 + H2 ),
(3.3)
O(H1 + 2H2 ), O(2H1 + 2H2 )i.
By the same argument, we obtain from (3.2) a similar decomposition

Db (Y ) = hLO(H2 ) C2 , O(H2 ), O(H1 ), O(H1 + H2 ), O(H1 + 2H2 ),
(3.4)
O(2H1 + H2 ), O(2H1 + 2H2 )i.
Up to permutation, the exceptional objects in the decompositions (3.3) and
(3.4) agree, hence they generate the same subcategory of Db (Y ). It follows
that LO(H1 ) C1 and LO(H2 ) C2 coincide, as both are the right orthogonal to the
same subcategory. Now the proposition follows since RO(H2 ) ◦ LO(H2 ) ∼ = id on
⊥
hO(H2 )i by Lemma 3.3.
Remark 3.11. The equivalence Db (X1 , α1 ) ' Db (X2 , α2 ) of Theorem 3.9

implies other relations between X1 and X2 . For instance, over k = C it
implies the Picard numbers of X1 and X2 agree. Indeed, it suffices to note that
the equivalence induces an isomorphism of twisted transcendental lattices
T(X1 , α1 ) ∼
= T(X2 , α2 ), whose ranks are the same as the usual transcendental
lattices (see [8]).
4. Equations for the twisted K3 surfaces and local invariants

Let k be a number field. Then for any place v of k, class field theory provides
an embedding invv : Br(kv ) → Q/Z (which is an isomorphism for nonarchi-
median v). Now let X be a smooth, projective, geometrically integral variety
over k. Any subset S ⊂ Br(X) cuts out a subset X(Ak )S ⊂ X(Ak ) of the
adelic points of X, given by
( )
X
S
X(Ak ) = (xv ) ∈ X(Ak ) | invv α(xv ) = 0 for all α ∈ S .
v
For fixed (xv ) ∈ X(Ak ) and α ∈ Br(X), the evaluation α(xv ) = 0 for all but
finitely many v, so the above sum is well-defined. Moreover, class field theory
gives inclusions
X(k) ⊂ X(Ak )S ⊂ X(Ak ).
Hence, if X(Ak )S is empty for some S, then X has no k-points. We note that
if X(Ak )S is empty but X(Ak ) is not, then S is said to give a Brauer–Manin
obstruction to the Hasse principle. See [12, 5.2] for more details.
In this section, we describe conditions on the divisor Z ⊂ P(V1 )×P(V2 )
from Section 2.2, which allow us to control the local invariants
invv α1 (xv ) for any v-adic point xv ∈ X1 (kv ). In the end, we will see that if
k = Q and enough conditions are met, then
(
0 if v is finite,
invv α1 (xv ) = 1
2 if v is real,
for all xv ∈ X1 (kv ). Hence X1 (Ak )α1 is empty and X1 has no k-points. Our
discussion follows [6] very closely, and differs only in the treatment of the
2-adic place (see Lemma 4.5).
Another random document with
no related content on Scribd:
across the clouds like spears of fire threatening revenge. The stone
has remained to the present day, and is known by the name of
“Koreno kardjo (dog) gambi” (stone).
PLATE XXXIV
The “Tjilbakuta” of the great emu ceremony, Arunndta tribe.

“The moment the sacred object is completed, the Illiyakuta delegates
one of his group to act as its attendant or guardian.”
The snake is an important character in the mythology of practically

every tribe of Australia; in fact most of the permanent water-holes
are supposed to be inhabited by great serpents which guard the
supplies, destroy unlawful consumers or polluters thereof, and
frequently communicate with those spirit ancestors of the tribes who
are descended from the original snake-man still living in the sky. In
many cases the mythic snakes can be recognized in some
characteristic features of the landscape. Take, for instance, the great
artesian spring near Coward Springs Station which is known as
Blanche Cup. This is looked upon as the mouth of a snake, while the
hill immediately at the back of it (Mt. Hamilton) is its head. In
consequence the formation is called “Worma-Kadiabba” (snake’s
head) by the local Arrabonna tribe. The natives have a dread of
these imaginary snake-monsters and prefer not to visit a water-hole
at night; in fact, at any time, day or night, they feel safer in the
company of a man who is “related” to the snake, because he can
protect them and give them the right of approach. The snake is
possessed of evil and will molest any but its totemic “relatives.”
The fundamental conception of the kobong (or totem), so far as
the Australian aboriginal is concerned, is of a religious nature. In the
beginning of all things, the Aluridja say a number of exalted
creatures of human form came out of the earth and were gracious to
their tribes-people. Then appeared a menace in the shape of a
gigantic dog which chased the good people from one place to
another, until they decided to adopt the forms of various animals and
plants, and thereby became either too fleet for the dog or were not
recognized by it. Other good people now descended from the hills
and drove the dog back to its hiding place in a cave where the evil
spirit dwells. The newcomers kindled a fire at the mouth of the cave
and kept the evil beings in captivity whilst the original Deities re-
assumed the human form. Ever after, however, these good creatures
were able to alter their appearance from human to animal at will; but
each individual in his choice adhered to the particular animal or plant
which had saved him from the ravages of the great evil dog.
Eventually they formed themselves into flat slabs of stone or wood,
upon the surfaces of which they scratched the emblems of their
animal representation and the traditions of their long wanderings on
earth. The spirits of these Deities now live in the sky but can return
at any time to re-enter the slab generally known as the “tjuringa.”
Among the Minning at Eucla the larger of these objects are known as
“wagal-wagal,” the smaller as “bobi,” whilst further west, in the
Laverton district, “kaidi” is the prevailing word. It is true, the tjuringa
is not known to all tribes; in which case the Deities are supposed to
have entered such natural objects as rocks, hills, and conspicuous
trees.
The Roper River natives believe that their deified forbears were
molested not by a dog, but by a hideous old woman or witch, who,
by the influence of evil, entrapped them and subsequently ate them.
On one occasion, however, a party of warriors were successful in
decoying her away from her haunts and slaying her. The jubilant
victors decided to cut out the old woman’s tongue as a trophy, but as
they were thus engaged, the tongue flew out of the mouth and spun
round in the atmosphere above them, making a terrible noise as it
did so. The men chased the tongue, but it flew towards a beefwood
tree and embedded itself deeply in the butt; in vain they looked for it
and tried to cut it out; it had become part of the tree. Before
returning, however, the men took a piece of wood out of the tree,
shaped like the woman’s tongue, which they tied to a piece of human
hair-string and swung round their heads with joy. Behold their mixed
feelings of delight and fear when the piece of wood began to howl
with a voice like that of the slain witch! The tribe retained that piece
of wood as a sacred memento of their victory, and they gave to it the
name the witch was known by, namely “Kunapippi.” Nowadays this
object is the equivalent of the central Australian tjuringa.
All tribes recognize the existence of deified ancestors, now real or
spiritual, whom they regard as sacred and worship accordingly. All
ancestors stand in a definite, intricate, and intimate relationship to
some animal, plant, water-hole, or other natural object which they
have at some time or other represented; some indeed in the first
place appeared as animals and later took the human form. They are
now looked upon as being those powers who by virtue of sacred
ceremonial can produce the species they have at some time
incarnated, in plenty or allow it to proliferate. As a matter of fact,
some of the sorcerers of the tribes often declare that they can see
the inside of a sacred rock or tjuringa teeming with young, ready to
be produced.
The Arunndta refer to their “Knaninja” (i.e. “totem” Deities) as
“Altjerrajara,” meaning the Supreme Number; the Aluridja as
“Tukurata” or “Tukutita”; and the Dieri as “Muramura.”
Just as the “totem” ancestor is connected with an animal, plant, or
other natural object, and is embodied in the sacred form of the
tjuringa, so the individual who traces his descent from such ancestor
recognizes a close and mysterious affinity between himself and the
tjuringa which has become his by heredity; henceforth it becomes his
sacred talisman which protects him from evil and procures for him
the means of maintaining his existence.
The emblematic representation of the deified ancestor, based
upon the form of an animal or plant living to-day and in some way
“connected” with the individual, is the “kobong” of the north-western
tribes first referred to by Sir George Grey.
The “totem” is very dear and sacred to the native, and is religiously
protected by him. I well remember on one occasion on the Alberga
River I discovered a small black and yellow banded snake which I
killed. An Aluridja man who was attached to the party at the time was
greatly shocked at this, and, with genuine sorrow, told me that I had
killed his “brother.” Turning to an Arunndta he lamented aloud:
“Kornye! Nanni kallye nuka kalla illum,” which literally translated
means: “Oh dear! This brother of mine is dead.”
One thing is always essential and that is that a native performs
frequent, prolonged, and reverential ceremonies, remote from the
women and children, and in the presence of his tjuringa. Under these
conditions the tjuringa is believed to have powers similar to those of
the Deity it embodies.
When not in use, the tjuringas are stored in caves, the entrances
to which are small and not easily discernible; the ground is
proclaimed taboo to any but initiated tjuringa holders and is strictly
regarded as a sanctum sanctorum. Although the sticks and stones
are the individual property of the tribesmen, the objects are generally
kept together, and only brought out during a religious ceremony. The
old men are the authorized custodians of the sacred collection. The
female tjuringas are included, because even though a woman may
possess one, she must never see it; if she does, accidentally or
otherwise, she is in imminent danger of being killed. No unauthorized
hunter is allowed near the prohibited area under any pretext at all;
even if an animal he has wounded should by accident make for the
sacred ground to breathe its last, the hunter is required by tribal law
and usage to sacrifice it to the divine factors incorporated in the
tjuringa, by leaving it on the spot.
PLATE XXXV
Flashlight photograph of “Illiya Tjuringa” or great emu ceremony, Arunndta

tribe.
“The chief emu man is distinguished by an extra large head-dress called the
‘Illiya Altjerra Kuta.’...”
When on the warpath, a warrior always craves to carry his tjuringa
with him, even though this is not always possible. He firmly believes
that with the talisman kept on his person, or at any rate knowing that
it is nearby, no deadly missile thrown by an enemy will penetrate his
body. The mere knowledge of the fact that his opponent has a
tjuringa with him, and he not, is sufficient to make a coward of the
bravest fighter. Should he be wounded or take ill, one of his “totem-
brothers” endeavours to produce a tjuringa, from which, if the
medicine man considers it necessary, a little powder is scraped and
handed to the patient to swallow with water.
With regard to “totem” animals which form the objects of hunting
expeditions, a man is allowed to kill and eat thereof with some
restriction. He must kill only one animal at a time, and only in
accordance with the method prescribed by the tribal fathers and
handed down to them by tradition. This is usually a straightforward
hunting method, with as little loss of blood as possible. If much blood
should flow, the hunter is obliged to cover it without delay with sand.
If possible, other men should cut up the carcase, and only certain
portions be handed to the “brother” of the slain animal.
Each tribe has an endless variety of objects (animal, vegetable,
terrestrial, meteoric, mythic, and so forth), which may figure as a
“totem.” Any one of these may be the primary motive of a separate
cult or sacred ceremony, but here again the variety is usually
reduced to the number corresponding to the most sanctimonious and
most useful creations affecting the affairs of the particular tribe.
The ceremonies take the form of either a direct worship or a
prayer for increased productivity of a certain plant or animal, either
being offered to the Knaninja or “totem” ancestors living as spirits in
the sky. Usually the two ideas are embodied in one grand ceremony,
and the method of procedure is governed by tradition. Such
ceremonies have been particularly elaborated by the Arunndta tribe,
who refer to them by the same name as that of the sacred object,
namely “Tjuringa”; less frequently they call them “Intitjuma,” the latter
name being applied more to ceremonies without worship.
The Tjuringa ceremonies are divided into grades according to their
importance and sacredness. The water ceremony is ordinarily called
“Kwatje Tjuringa,” but if the “totem” spirit ancestor is invoked to
attend, it goes by the name of “Kwatje Tjuringa Knaninja”; if the
principal spirit ancestor is assumed to be present, the title becomes
“Kwatje Tjuringa Knaninja Knurrendora”; and finally the most sacred
water ceremony of all is the “Kwatje Tjuringa Altjerra Knaninja
Knurrendora.”
As a typical illustration we shall discuss the “Illiya Tjuringa” or
Great Emu Ceremony of the eastern Arunndta groups. The date of
the performance is decided by the senior emu “brother” of the tribe,
the oldest member who claims to be related to the Illiya Knaninja.
Somewhat extensive preparations are made beginning a few days
prior to the opening event. Only fully initiated men take part, but the
women are allowed to witness certain of the most awe-inspiring
stages from a distance. Whilst the younger men are out collecting
leaves, out of which they make the down later to adorn the bodies of
the performers, the older men prepare the sacred ground. Others
slay a number of brown hawks, off which they pull the feathers and
then pluck the down. A suitable site having been selected, the old
men clear it by removing all grass and bush from the surface and
smoothing the sand with their feet. The “brothers” who claim
relationship alike to the great Emu-Man, the Emu-Knaninja, and the
emu itself, thereupon proceed to anoint the sacred ground with their
blood, for which purpose they puncture the median basilic vein of the
forearm with a quartzite chip and allow the fluid of kinship to sprinkle
upon the sand. It is surprising to see the amount of blood sacrificed
by the men on occasions like this; and time after time, when such is
required, the process is repeated. By examining the forearms of an
old stager, one can usually count a number of small scars along the
course of a vein indicating places where a perforation has at different
times been made. A venesection is made after much the same
manner among the various tribes.
The following Arunndta method will serve as an example. A
ligature of hair-string is in the first place tied tightly round the upper
arm, a little above the biceps muscle, after the style of a tourniquet to
check the flow of blood in the veins and thereby distend the vessels.
The man then makes a small longitudinal cut through the skin and
punctures the vein beneath it lengthwise; the blood spurts forth
immediately and is collected in the handle-pit of a shield. When the
flow is to be stopped, the native removes the ligature, and this in
most cases is all that is needed. Should, however, the blood
continue to come, he places a small amount of down over the
incision and presses it against the vein, or winds three or four
strands of fur-string around it. The little pad of down is usually left on
the arm until it dries and falls off. None of the women are allowed to
witness this operation, which is called “Ilgarukna.” The blood, when it
is to be used as an adhesive for the down-decoration, is applied with
a small brush (“ipinja”) made of twigs tied together with fur-string.
Vide Plate XXXIII.
The principal among the emu group is called “Illiyakuta,” and it is
he who directs the performance. He takes his followers to a secluded
place, such as a clump of timber or down a creek-bed, and there the
wooden tjuringas belonging to the ceremony are produced and
painted afresh with red ochre and emu fat.
Down is made out of the white, felty leaves and twigs of Kochia
bush, which the Arunndta call “kemba.” Small quantities of these are
placed upon a flat slab of stone and pounded with a pebble. The
fluffy material which results is next mixed and rubbed by hand with
powdered kaolin or ochre according to the colour required, the white
being known as “wadua,” the red as “wanjerra.”
A sacred object is now constructed which encloses the painted
tjuringas and is called the “Tjilbakuta.” It is about three feet high and
is made in the following way. The tjuringas are laid one on top of the
other and bound together with many lengths of human hair-string,
which completely obscure the shape of the separate pieces. A thick
layer of the stalks of the kangaroo grass (Anthistiria) is laid around
the parcel and kept in position with a few lengths of twine, and then
the whole structure is covered with great masses of human hair-
string wound spirally from top to bottom. A cylinder results which is
decorated with alternate vertical bands of red and white vegetable
down. Into the top of this Tjilbakuta one bundle of emu feathers and
one of black cockatoo tail-feathers are stuck; and often additional
plumes are hung beneath them. The moment the sacred object is
completed, the Illiyakuta delegates one of his group to act as its
attendant or guardian. For the time being his body is decorated with
symmetrically placed, curved ochre bands upon the chest and
vertical bands down the arms; at a later stage he ornaments his
body more elaborately, prior to taking part in the principal
performance; but all the time he remains in his place of hiding beside
the Tjilbakuta. Vide Plate XXXIV.
At the sanctified place close by the other men have been stacking
firewood at different points to illuminate the proceedings during the
evening. Occasionally, too, the Illiyakuta group of men cover a
portion of the ground with a coloured emblem of the traditional emu.
Early in the afternoon of the festive day the men who will take part
in the ceremony at night begin to prepare themselves. Many of the
non-performers help them.
Large quantities of down, both vegetable and birds’, are used to
decorate the bodies. The design is shaped much like a cobbler’s
apron, extending from the neck down the front to the level of the
knees. The greater part of this surface is red, but it is lined with white
and split along the centre by two parallel lines of white. The back is
not decorated at all. The entire surface of the face, including the eye-
lids and beard, is thickly covered with down which is white, except
for an oval red patch around the mouth.
The principal attraction, however, of the sacred emu ceremony is
the head-dress, which is both elaborate and imposing. To prepare it,
the attendant combs back the actor’s hair with his fingers, and
interlaces it with stalks of grass and small twigs in such a way that a
tall conical structure results right on top of the head. This is made
secure and of a uniform exterior by winding much human hair-string
around it, at the same time taking in a plume of emu feathers at the
apex of the cone. The headgear is completely enveloped in red and
white down, extending upwards from the head as alternate vertical
bands. The chief emu-man is distinguished by an extra large
headdress called the “Illiya Altjerra Kuta”; this measures a good
three feet in length, and it embodies, between the apex and the emu
plume, deeply enshrouded with hair-string and down, the sacred
“Illiya Tjuringa.” Other members who are of the same rank as the
“Tjilbakuta” guardian, wear their insignia beneath the emu feathers in
the form of a sickle-shaped rod, which carries at each of its points a
tuft of white cockatoo feathers. All performers cover their person with
a dog-tail appendage which hangs from a thin waistband of human
hair-string. And lastly, they all tie bundles of eucalyptus twigs, with
the leaves attached, to their legs just above the ankles. If possible,
old or half-dried leaves are selected in order that a more pronounced
rustling is produced when the men move about; the noise is made to
imitate the rustle of the wiry feathers of an emu. Vide Plate XXXV.
At nightfall the Tjilbakuta is removed from the hiding place and
planted on the edge of the ensanguined patch. The guardian is thus
given an opportunity to slip away and to attend to his ceremonial
toilet, which is similar to that of the rest of the Tjilbakuta group. When
he returns, the performance is about to begin, and all except he
leave the ground.
The stacks of wood are set fire to by invisible hands, and, so soon
as the flames flare upwards, the silence is broken by the booming
note of a bull-roarer, which is produced some distance off in the
bush.
The Tjilbakuta guardian sits beside the object like a statue, with
his eyes rivetted to the ground immediately in front of him. From
behind him the thud of stamping feet and the rustle of dry leaves
announce the coming of the official performers, while from the other
side the non-performing members step from the darkness and take
up their position by squatting between two fires. When the decorated
men come into view, the latter start beating their boomerangs
together in perfect time to the stamping of the feet of the advancing
actors. They come as a body of five or six rows, one behind the
other, each man holding his hands locked behind his back and
uttering a deep guttural note resembling a pig’s grunt. The folded
hands held over the stern represent the tail, the guttural noise the
call of the emu.
The Illiyakuta, wearing the tall Illiya Altjerra Kuta, is in the front
row, and he is attended on either side by a Tjilbakuta man. The chief
now starts a chant: “Immara janki darrai,” and all the others,
including the sitting men, join in; the same is repeated several times.
When the two parties are opposite each other, the performers
quicken the pace of their stamping and extend their arms sideways,
thereby widening their ranks. After this they retreat to behind the
Tjilbakuta and one hears a shrill chirping note resembling the cry of a
young emu.
The interpretation of this act needs no special elucidation. The
decorated performers are those of the tribe’s manhood who, in all
matters pertaining to the emu, have a right to communicate, through
the Tjilbakuta, with the astral emu ancestor living in the great
celestial domain of the ancestral spirits, which is known as
“Altjerringa.” They are invoking the benign Knaninja or originator of
their particular “totem” species to increase the numbers of emu on
earth for the exclusive benefit of their tribe. It is the Illiyakuta who
imagines that he receives the favourable response from above, and,
when it comes, it is he who imitates the cry of a young emu. It often
happens, however, that the chief persuades himself to believe that
the Great Spirit had not heeded the appeal, in which case the last-
mentioned cry is wanting. The ceremony is repeated time after time.
Altjerringa, it will be observed, is a compound word consisting of
“Altjerra,” the Supreme Spirit, and “inga,” a foot or trail. The implied
idea is that Altjerringa is the “walk-about” of the spirit ancestors,
where they walk, and have always walked, and where the spirits of
all tribes-people eventually hope to find their way.
After this act, the performance becomes less restrained and takes
more the form of a corrobboree. Some of the men seize firebrands
from the burning stacks and hurl them in the direction of the
women’s camp. From the moment of the sounding of the bull-roarer
at the beginning of the ceremony until now the women sat huddled
together, with their faces buried in their hands, thoroughly cowed by
the portentous happenings. When the firebrands come whizzing
through the air and crash into the branches of the trees around them,
sending sparks flying in all directions, they are almost beyond
themselves with fear. But just at this juncture the men call upon them
to look towards the festive ground and behold them dancing. In
obedience to the order, the women’s fears are dispelled and soon
superseded by a noticeable enravishment. They feast their eyes
upon the array of manhood in gala dress, and it is not long ere they
pick up the rhythm of a dance by beating time to the step. Provided
the Tjilbakuta has been removed to a place of secrecy, well out of
reach of accidental discovery, the men entreat the women to come
up and join in the song. Thus the sublime is eventually reduced to
commonplace, and the remainder of the night passes in joviality.
To refer briefly to a vegetable ceremony, we shall select the yam
or “Ladjia Tjuringa Knaninja.” The preparations are much the same
as those of the emu ceremony. An enclosure is first made in a
secluded spot with branches, in the centre of which the “totem” or
Knaninja “stick” is erected. Several men immediately set about to
decorate it with vegetable down as previously described. The design
in this case consists of vertical rows of red circles upon a yellow
ochre background. In addition, a large plume of split eagle-hawk
feathers is stuck into the top of the stick. All ordinary performers
wear conical head-gears or “tdela” made of Cassia twigs, into the
apices of which tightly bound bundles of grass stalks (“gortara”) are
fixed carrying plumes of emu feathers (“mangalingala”) (Plate
XXXVI, 1). Other men have squat, cylindrical bark structures called
“elbola” placed over their heads, which are elaborately decorated
with vertical coils of human hair-string and coloured down.
One of the principal actors represents the “Kuta Knaninja.” His
head-gear consists of two long kutturu, tied together with hair-string
and completely covered with gum leaves, the whole being
subsequently besmeared with blood and decorated with coloured
down. As the assistants are dressing this character, they keep up a
chant sounding like “Winni kutcherai.” Vide Plate XXXVI, 2.
The leading figure is the “Ingada Ladjia Knaninja,” who wears a tall
vertical head-piece which contains the tjuringa of the Ladjia
Knaninja. The tjuringa is, however, not visible, but is covered with
pieces of bark, securely tied over it with hair-string, the whole being
richly decorated with vertical bands of red and white down.
The Great Spirit of the Yam, called “Knaninja Tjilba Ladjia,” when
he leaves Altjerringa, takes up his abode in a cave near Mount
Conway, where the tjuringas are kept, but at night, before the fires
are lit, he is supposed to come to the ceremonial ground and occupy
the decorated “totem” stick described above. During the performance
he is surrounded by all the ordinary performers, who are known as
“Tjilba Ingarrega,” and are directly under the guidance of the Ingada
and Kuta Knaninja.
A group of men who are not decorated sit near one of the fires and
sing while the performers are thus encircling the Ladjia stick:
“Imbanai yinga
Wi ma bana Ladji di bana
Yammana wi ma bana
Jai ra ja ja
Jai ja ja na
Wi ba na na
De a re a ja betja,”
the voices finally fading away to an almost inaudible whisper.

If the Great Spirit, Ladjia Altjerra Knaninja, is gracious, the tap-root
of the yam will be sent deep down into the earth near the Jay River
and from there spread its laterals all over the country to supply the
needs of the tribe.
When some of the most sacred ceremonies are performed, the
oldest “relatives” of the presiding Knaninja often construct a coloured
drawing upon the consecrated ground, whose purpose is similar to
that of the “totem” stick above described. The drawing is executed in
coloured down, both vegetable and bird. A space of suitable size,
often measuring many feet in length, is cleared of grass and stones,
and sprinkled with water, when it is ready to receive the down. In the
case of, say, the “Ladjia Tjuringa,” the design takes the form of a
number of concentric circles alternately red and white, from the
outermost of which six equally spaced groups of red and white lines
stand out radially. The enclosing border of the design consists
entirely of white down. Vide Plate XXXVII.
Once constructed, this drawing, which is known as “Etominja,” is
zealously guarded by one of the old men. If, peradventure, an
unauthorized person happens upon the sanctified place, he is killed
and buried immediately beneath the spot occupied by the design;
thereupon the ground is smoothed again and the Etominja re-
constructed. Nobody in camp ever hears what became of the
person, and should any relative track him in the direction of the area
known to be tabooed, he is horror-stricken and runs away.
While the old men are re-constructing the Etominja, they sing to
the Knaninja as follows:
“Yedimidimi
Dakarai pa ma taka,
Pa mitu min jai, jin tarai,
Ja ra nai malgada, ja ranai.”
The next, and probably the most important, group of religious

ceremonies is that dealing with Sex-Worship. For years past
peculiarly shaped stones have been found in caves and among the
possessions of the Australian aborigines whose shape was strikingly
suggestive of a phallus, but hitherto no actual phallic ceremonies
have been observed. It was my good fortune to witness such among
the Aluridja, Arunndta, Dieri, and Cambridge Gulf tribes. From
enquiries made of the old men, it appears that in former days this
form of worship was practised considerably more than it is
nowadays. New stone phallus are rarely made by the present tribes;
those in their possession have generally been inherited from
previous generations. The old men have the phallus in their keeping,
and they are very loth to either produce or part with them.
The natives of the King Sound district in the north-west believe the
origin of the phallus to be as follows: In the early times a scourge
was raging among their forefathers, from the effects of which many
were daily dying, when a hairy man and his mate, a woman of
ordinary human form, came to earth from above. The evil was due to
the exhalation of poisonous breath from the gaping jaws of a green
monster resembling a crocodile. The stranger relieved the sufferers
from the awful curse by showing them how to perform an operation
upon their person which taught them to endure pain and protected
them against future ravages of the pestilence. This great and
benevolent stranger then took his departure, but left his name to
designate the surgical operation which to the present day is
performed upon the male members of the tribe; the name, strange
though it may seem, is “Elaija”; and it is known, at any rate, as far
east as Port George IV. But the tribe had become so weak through
the terrible havoc the disease had wrought that the old men called
him back and entreated him to stay. Elaija, however, took from a
dillybag his female companion was carrying, a stone carved after the
shape of a mutilated member, which he gave the name of
“Kadabba.” When the old men gazed upon this object, they took
fright and appealed to Elaija, but the good fellow had vanished. The
stone has remained with the tribe ever since, and through the divine
property Elaija endowed it with, their threatened extinction was
eluded. Moreover, they continued to practise the operation on all
young men because it made their members like the Kadabba of
Elaija, which they knew had the power of multiplying their kind. And
so the Kadabba became a sacred object whose procreative power
they have learned to worship, thinking that by such observance they
would augment their own capacities of sex. Vide Fig. 7.
Fig. 7. Stone phallus, Northern Kimberleys, Western Australia (× 1/2).
One often reads, and I was under the same impression myself
until I became better acquainted with the tribes, that the Australian
natives do not connect the knowledge of conception with any
intercourse which might have taken place between the sexes. This I
find is not altogether correct, although usually the younger people
are kept in complete ignorance on the subject. No doubt strangers
are treated similarly when they put any pertinent questions to the old
men on matters of sex. The old men believe in the duality of human
creation, the spiritual and the material; sexuality is regarded as the
stimulus of corporeal reproduction, but the spirit quantity is derived
through mystic and abstract influences controlled by a “totem”-spirit
or Knaninja. Under these circumstances, it is not surprising to note
that the ceremonies of the phallus are transacted principally by the
old men of the tribe who aim at the rejuvenation of their waning
powers.
It is interesting to see the old men preparing for a ceremony which
is to be dedicated to a Knaninja or Spirit of Sex, because they all
endeavour to conceal the white hairs of their beards by rubbing
powdered charcoal into them. The bark of the cork tree (Hakea) is
used for the purpose; pieces of it are charred, crushed between the
palms, and applied where needed. It is astounding what a difference
this process makes to the appearance, and some of the old grey-
beards really look as though they had been made twenty years
younger by magic.
In the eastern MacDonnell Ranges stands a cylindro-conical
monolith whose origin is believed to be as follows: Many generations
ago, the paternal ancestors of the Arunndta walked from a district
situated, as near as one can gather, somewhere in the
neighbourhood of Ediowie; they were known as the “Kukadja,” and
were characterized by the enormous dimensions of their organs.
These old men or Tjilba of the tribe migrated northwards to beyond
Tennant’s Creek and settled in the productive “Allaia” country which
surrounds the Victoria River. In that same district one finds, even at
the present day, cave drawings of human beings with the anatomical
peculiarities referred to (Fig. 8). At a later time, the head-man of the
Kukadja, named “Knurriga Tjilba,” returned southwards to the
Macdonnell Ranges. While roaming the hills, he espied two young
women sitting on the side of a quartzite cliff, and without deliberation
began to approach them. He was in the act of making lewd overtures
when the guardian of the girls, a crow ancestor, caught sight of him
and hurled a boomerang at him. The missile struck the great man
and cut off the prominent portion of his body, which in falling stuck
erect in the ground. The force of the impact was so great that the
man bounced off the earth and fell somewhere near Barrow’s Creek.
He bled so profusely that a clay-pan soon filled with his blood. Thus
his followers found him, and overcome with sorrow they opened the
veins of their arms to mix their blood with his. Then all the members
of the party jumped into the pool and disappeared for ever.
Fig. 8. Ochre drawing of “Kukadja” men, north of Wickham River, Northern
Territory (× 1/3).
The severed portion of the old man’s body, however, remained just
where it fell and turned to stone. It has long been known as “Knurriga
Tjilba Purra.”
The two young women can also still be detected in the cliff as
prominent rock formations.
The stone has been protected by the tribe as long as the old men
can remember, because they realize that it contains an inexhaustible
number of unborn tribes-people. These mythic, foetal elements are
generally recognized to exist in certain objects of phallic significance,
and are called “rattappa.” The medicine men maintain that they can
at times see the dormant living matter in the stone. It is on that
account that it is regarded as sacred, and every now and then very
secret and worshipful ceremonies are transacted near its base, the
main objects of which are to multiply the future membership of the
tribe and to preserve the sexual powers of the old men.
The Tjilba Purra naturally figures prominently in some of their
ceremonies. In fact, it is reproduced and worn upon the head of the
leading man during the functions. The sacred effigy consists of an
upright column, about two feet high, composed of a stout bundle of
grass stalks, in the centre of which the tjuringa is contained. It is
decorated with alternating bands of red and white down throughout
its length. This upright column represents the “Tjilba” or revered
ancestor whose spirit is invoked to “sit” in the tjuringa; at the top of it
a plume of wiry emu feathers, well powdered with charcoal (“unjia”)
to give it a youthful appearance, takes the place of the forbear’s hair
and beard. Standing at an angle with the central column, a similar
though slightly smaller structure is intended for the “Purra” or
phallus; it carries a plume of white cockatoo feathers at its end to
represent the glans. Vide Plate XXXVIII, 2.
A landmark, of similar significance as the Tjilba Purra of the
Arunndta, exists on the Roper River in the Northern Territory; it is a
pillar of sandstone known as “Waraka.” Waraka is also the name of
the great Spirit Father of the tribe. In very early times this man came
to earth in a semi-human form, and made the country abound in
game, animals, birds, and fish. Then he found a woman on the
shores of Carpentaria Gulf who remained with him as his wife. Many
children came of the union; and Waraka’s mate has since been
looked upon as the mother of the tribe. The woman’s name was
“Imboromba,” and to this day the tribe takes its name after her.
Warraka had an enormous sex characteristic which was so
ponderous that he was obliged to carry it over one of his shoulders.
Eventually the organ became so huge that Warraka collapsed and
sank into the earth. His burden remained, but turned to stone, and is
now looked upon by the local natives as the great symbol of Nature’s
generative power which first produced their game supplies and then
the original children of the tribe; it is revered accordingly.
The Kukata have a somewhat similar legend of the origin of a
stone of phallic significance, the name of the possessor of the large
organ being “Kalunuinti.”
In the extreme north-western corner of Australia, in the Glenelg
River district, the natural stone is replaced by an artificially

Textbook Brauer Groups and Obstruction Problems Moduli Spaces and Arithmetic 1St Edition Asher Auel Ebook All Chapter PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Textbook Brauer Groups and Obstruction Problems Moduli Spaces and Arithmetic 1St Edition Asher Auel Ebook All Chapter PDF

Uploaded by

Copyright:

Available Formats

Brauer Groups and Obstruction

Problems Moduli Spaces and

Algebra 1 Groups Rings Fields and Arithmetic 1st

A Primer on Hilbert Space: Theory Linear Spaces,

A Primer on Hilbert Space Theory Linear Spaces

A Gentle Course in Local Class Field Theory Local

Found and Forged 2nd Edition Ivy Asher

Theory of Groups and Symmetries Finite groups Lie

Aspects of Scattering Amplitudes and Moduli Space

More information about this series at http://www.springer.com/series/4848

ISSN 0743-1643 ISSN 2296-505X (electronic)

© Springer International Publishing AG 2017

Printed on acid-free paper

This book is published under the trade name Birkhäuser, www.birkhauser-science.com

Kenneth Ascher, Krishna Dasaratha, Alexander Perry, and Rong Zhou

Asher Auel, Jean-Louis Colliot-Thélène, and Raman Parimala

Fedor Bogomolov and Yuri Tschinkel

Brendan Hassett and Yuri Tschinkel

Colin Ingalls, Andrew Obus, Ekin Ozman, and Bianca Viray

Martı́ Lahoz, Emanuele Macrı̀, and Paolo Stellari

Kelly McKinnie, Justin Sawon, Sho Tanimoto, and

Let S1 and S2 denote complex K3 surfaces. An isomorphism

classes on a degree 2 K3 surface with Picard rank 1. The paper of McKinnie,

establishes a local-global criterion for the decomposability of elements in de-

Abstract. In this short note we observe that the recent examples of

Some years ago Gross and Popescu [12] studied a simply-connected

© Springer International Publishing AG 2017 1

Ψ : Db (Y ) → Db (X) are inverse equivalences, then by [20, Thm. 2.2] there

[22] M. Popa and C. Schnell. Derived invariance of the number of holomorphic 1-

Abstract. We show how work of Rickard and Toën completely resolves

© Springer International Publishing AG 2017 7

2. Twisted derived categories

If X is a scheme and α ∈ Hét 2

3. Twisted derived equivalences over affine schemes

3. there is an R-linear equivalence of triangulated categories Db (R, α) '

Remark 3.3. By [3], Br(R) = Br0 (R) = Hét

of a scheme classifies stacks of locally presentable dg categories that are étale

We end by observing that the condition of R-linearity in Theorem 3.2 is

Abstract. Using a construction of Hassett and Várilly-Alvarado, we pro-

© Springer International Publishing AG 2017 13

2. Construction of the twisted K3 surfaces

2.1. Quadric fibrations

2.2. Twisted K3 surfaces

Proof. Consider the case i = 1. The morphism j1 : Y → P(E1 ) is given

3. Derived equivalence of the twisted K3 surfaces

to the derived category of a point via Db (Spec(k)) → Db (X) : V 7→ V ⊗F. To

3.3. Derived categories of quadric fibrations

Proof. Set Ci = Φi (Db (P(Vi ), C`0,i )) ⊂ Db (Y ). Theorem 3.6 gives semior-

Recall Beilinson’s decomposition Db (P(Vi )) = hO, O(Hi ), O(2Hi )i (see Ex-

By the same argument, we obtain from (3.2) a similar decomposition

Remark 3.11. The equivalence Db (X1 , α1 ) ' Db (X2 , α2 ) of Theorem 3.9

4. Equations for the twisted K3 surfaces and local invariants

The “Tjilbakuta” of the great emu ceremony, Arunndta tribe.

The snake is an important character in the mythology of practically

Flashlight photograph of “Illiya Tjuringa” or great emu ceremony, Arunndta

the voices finally fading away to an almost inaudible whisper.

The next, and probably the most important, group of religious

You might also like