Download the Full Version of textbook for Fast Typing at textbookfull.

com

An Introduction to Description Logic 1st Edition

Franz Baader

https://textbookfull.com/product/an-introduction-to-
description-logic-1st-edition-franz-baader/

OR CLICK BUTTON

DOWNLOAD NOW

Download More textbook Instantly Today - Get Yours Now at textbookfull.com

 Recommended digital products (PDF, EPUB, MOBI) that
you can download immediately if you are interested.

An Introduction to Formal Logic 2nd Edition Peter Smith

https://textbookfull.com/product/an-introduction-to-formal-logic-2nd-
edition-peter-smith/

textboxfull.com

Introduction to Logic Programming 1st Edition Michael

Genesereth

https://textbookfull.com/product/introduction-to-logic-
programming-1st-edition-michael-genesereth/

textboxfull.com

Introduction to Logic 3rd Edition Michael Genesereth

https://textbookfull.com/product/introduction-to-logic-3rd-edition-
michael-genesereth/

textboxfull.com

Logic As A Liberal Art An Introduction To Rhetoric And

Reasoning Rollen Edward Houser

https://textbookfull.com/product/logic-as-a-liberal-art-an-
introduction-to-rhetoric-and-reasoning-rollen-edward-houser/

textboxfull.com
Introduction to Logic Third Edition Michael R Genesereth

https://textbookfull.com/product/introduction-to-logic-third-edition-
michael-r-genesereth/

textboxfull.com

Introduction to Logic 14th Edition Irving M. Copi

https://textbookfull.com/product/introduction-to-logic-14th-edition-
irving-m-copi/

textboxfull.com

Introduction to Formal Logic 2nd Edition Russell Marcus

https://textbookfull.com/product/introduction-to-formal-logic-2nd-
edition-russell-marcus/

textboxfull.com

Introduction to Category Theory and Categorical Logic

Thomas Streicher

https://textbookfull.com/product/introduction-to-category-theory-and-
categorical-logic-thomas-streicher/

textboxfull.com

Processing an introduction to programming 1st Edition

Nyhoff

https://textbookfull.com/product/processing-an-introduction-to-
programming-1st-edition-nyhoff/

textboxfull.com
 "O *OUSPEVDUJPO UP %FTDSJQUJPO -PHJD

%FTDSJQUJPO MPHJD %- IBT B MPOH USBEJUJPO JO DPNQVUFS TDJFODF BOE LOPXMFEHF

SFQSFTFOUBUJPO CFJOH EFTJHOFE TP UIBU EPNBJO LOPXMFEHF DBO CF EFTDSJCFE BOE TP
UIBU DPNQVUFST DBO SFBTPO BCPVU UIJT LOPXMFEHF %- IBT SFDFOUMZ HBJOFE JODSFBTFE
JNQPSUBODF TJODF JU GPSNT UIF MPHJDBM CBTJT PG XJEFMZ VTFE POUPMPHZ MBOHVBHFT JO
QBSUJDVMBS UIF XFC POUPMPHZ MBOHVBHF 08-
8SJUUFO CZ GPVS SFOPXOFE FYQFSUT UIJT JT UIF àSTU UFYUCPPL PO %FTDSJQUJPO -PHJD
*U JT TVJUBCMF GPS TFMGTUVEZ CZ HSBEVBUFT BOE BT UIF CBTJT GPS B VOJWFSTJUZ DPVSTF
4UBSUJOH GSPN B CBTJD %- UIF CPPL JOUSPEVDFT UIF SFBEFS UP JUT TZOUBY TFNBOUJDT
SFBTPOJOH QSPCMFNT BOE NPEFM UIFPSZ BOE EJTDVTTFT UIF DPNQVUBUJPOBM DPNQMFYJUZ
PG UIFTF SFBTPOJOH QSPCMFNT BOE BMHPSJUINT UP TPMWF UIFN *U UIFO FYQMPSFT B WBSJFUZ
PG EJGGFSFOU EFTDSJQUJPO MPHJDT SFBTPOJOH UFDIOJRVFT LOPXMFEHFCBTFE BQQMJDBUJPOT
BOE UPPMT BOE EFTDSJCFT UIF SFMBUJPOTIJQ CFUXFFO %-T BOE 08-
"O *OUSPEVDUJPO UP %FTDSJQUJPO -PHJD

'3"/; #""%&3
5FDIOJTDIF 6OJWFSTJUÅU %SFTEFO

*"/ )0330$,4
6OJWFSTJUZ PG 0YGPSE

$"345&/ -65;
6OJWFSTJUÅU #SFNFO

6-* 4"55-&3
6OJWFSTJUZ PG .BODIFTUFS
6OJWFSTJUZ 1SJOUJOH )PVTF $BNCSJEHF $# #4 6OJUFE ,JOHEPN
0OF -JCFSUZ 1MB[B UI 'MPPS /FX :PSL /:  64"
 8JMMJBNTUPXO 3PBE 1PSU .FMCPVSOF 7*$  "VTUSBMJB
 OE 'MPPS "OTBSJ 3PBE %BSZBHBOK %FMIJ m  *OEJB
 "OTPO 3PBE  4JOHBQPSF 

$BNCSJEHF 6OJWFSTJUZ 1SFTT JT QBSU PG UIF 6OJWFSTJUZ PG $BNCSJEHF

*U GVSUIFST UIF 6OJWFSTJUZT NJTTJPO CZ EJTTFNJOBUJOH LOPXMFEHF JO UIF QVSTVJU PG
FEVDBUJPO MFBSOJOH BOE SFTFBSDI BU UIF IJHIFTU JOUFSOBUJPOBM MFWFMT PG FYDFMMFODF

XXXDBNCSJEHFPSH
*OGPSNBUJPO PO UIJT UJUMF XXXDBNCSJEHFPSH
%0* 
¥ 'SBO[ #BBEFS *BO )PSSPDLT $BSTUFO -VU[ BOE 6MJ 4BUUMFS 
5IJT QVCMJDBUJPO JT JO DPQZSJHIU 4VCKFDU UP TUBUVUPSZ FYDFQUJPO
BOE UP UIF QSPWJTJPOT PG SFMFWBOU DPMMFDUJWF MJDFOTJOH BHSFFNFOUT
OP SFQSPEVDUJPO PG BOZ QBSU NBZ UBLF QMBDF XJUIPVU UIF XSJUUFO
QFSNJTTJPO PG $BNCSJEHF 6OJWFSTJUZ 1SFTT
'JSTU QVCMJTIFE 
1SJOUFE JO UIF 6OJUFE ,JOHEPN CZ $MBZT 4U *WFT QMD
" DBUBMPHVF SFDPSE GPS UIJT QVCMJDBUJPO JT BWBJMBCMF GSPN UIF #SJUJTI -JCSBSZ
*4#/  )BSECBDL
*4#/  1BQFSCBDL
$BNCSJEHF 6OJWFSTJUZ 1SFTT IBT OP SFTQPOTJCJMJUZ GPS UIF QFSTJTUFODF PS BDDVSBDZ
PG 63-T GPS FYUFSOBM PS UIJSEQBSUZ *OUFSOFU 8FC TJUFT SFGFSSFE UP JO UIJT QVCMJDBUJPO
BOE EPFT OPU HVBSBOUFF UIBU BOZ DPOUFOU PO TVDI 8FC TJUFT JT PS XJMM SFNBJO
BDDVSBUF PS BQQSPQSJBUF
 Contents

1 Introduction page 1
1.1 What are DLs and where do they come from? 1
1.2 What are they good for and how are they used? 3
1.3 A brief history of description logic 4
1.4 How to use this book 7
2 A Basic Description Logic 10
2.1 The concept language of the DL ALC 10
2.2 ALC knowledge bases 16
2.2.1 ALC TBoxes 17
2.2.2 ALC ABoxes 19
2.2.3 Restricted TBoxes and concept definitions 23
2.3 Basic reasoning problems and services 28
2.4 Using reasoning services 36
2.5 Extensions of the basic DL ALC 37
2.5.1 Inverse roles 37
2.5.2 Number restrictions 39
2.5.3 Nominals 41
2.5.4 Role hierarchies 42
2.5.5 Transitive roles 42
2.6 DLs and other logics 44
2.6.1 DLs as decidable fragments of first-order logic 44
2.6.2 DLs as cousins of modal logic 46
2.7 Historical context and literature review 48
3 A Little Bit of Model Theory 50
3.1 Bisimulation 51
3.2 Expressive power 53
3.3 Closure under disjoint union 55

v
vi Contents
3.4 Finite model property 57
3.5 Tree model property 63
3.6 Historical context and literature review 67
4 Reasoning in DLs with Tableau Algorithms 69
4.1 Tableau basics 70
4.2 A tableau algorithm for ALC 71
4.2.1 ABox consistency 72
4.2.2 Acyclic knowledge base consistency 82
4.2.3 General knowledge base consistency 83
4.3 A tableau algorithm for ALCIN 90
4.3.1 Inverse roles 90
4.3.2 Number restrictions 93
4.3.3 Combining inverse roles and number restrictions 97
4.4 Some implementation issues 101
4.4.1 Or-branching 101
4.4.2 And-branching 103
4.4.3 Classification 104
4.5 Historical context and literature review 104
5 Complexity 106
5.1 Concept satisfiability in ALC 107
5.1.1 Acyclic TBoxes and no TBoxes 108
5.1.2 General TBoxes 117
5.2 Concept satisfiability beyond ALC 123
5.2.1 ALC with inverse roles and nominals 123
5.2.2 Further adding number restrictions 125
5.3 Undecidable extensions of ALC 130
5.3.1 Role value maps 130
5.3.2 Concrete domains 134
5.4 Historical context and literature review 137
6 Reasoning in the EL Family of Description Logics 140
6.1 Subsumption in EL 141
6.1.1 Normalisation 142
6.1.2 The classification procedure 147
6.2 Subsumption in ELI 151
6.2.1 Normalisation 151
6.2.2 The classification procedure 152
6.3 Comparing the two subsumption algorithms 159
6.3.1 Comparing the classification rules 159
6.3.2 A more abstract point of view 162
 Contents vii
6.4 Historical context and literature review 165
7 Query Answering 168
7.1 Conjunctive queries and FO queries 169
7.2 FO-rewritability and DL-Lite 174
7.2.1 Introducing DL-Lite 175
7.2.2 Universal models 180
7.2.3 FO-rewritability in DL-Lite 184
7.3 Datalog-rewritability in EL and ELI 192
7.3.1 Fundamentals of Datalog 193
7.3.2 Datalog-rewritings in ELI 195
7.3.3 Short Datalog-rewritings in EL 198
7.4 Complexity aspects 199
7.5 Historical context and literature review 202
8 Ontology Languages and Applications 205
8.1 The OWL ontology language 206
8.1.1 OWL and RDF 206
8.1.2 OWL and SROIQ 209
8.1.3 OWL ontologies 213
8.1.4 Non-DL features 217
8.1.5 OWL profiles 222
8.2 OWL tools and applications 223
8.2.1 The OWL API 223
8.2.2 OWL reasoners 224
8.2.3 Ontology engineering tools 224
8.2.4 OWL applications 225
Appendix: Description Logic Terminology 228
A.1 Syntax and semantics of concept and role constructors 228
A.2 Syntax and semantics of knowledge bases 230
A.3 Naming schemes for description logics 231
References 234
Index 252
 1
Introduction

This is, to the best of our knowledge, the first textbook dedicated solely
to Description Logic (DL), a very active research area in logic-based
knowledge representation and reasoning that goes back to the late 1980s
and that has a wide range of applications in knowledge-intensive infor-
mation systems. In this introductory chapter we will sketch what DLs
are, how they are used and where they come from historically. We will
also explain how to use this book.

1.1 What are DLs and where do they come from?

Description logics (DLs) are a family of knowledge representation lan-
guages that can be used to represent knowledge of an application domain
in a structured and well-understood way.1 The name description logics
is motivated by the fact that, on the one hand, the important notions
of the domain are represented by concept descriptions, i.e., expressions
that are built from atomic concepts (unary predicates) and atomic roles
(binary predicates) using the concept and role constructors provided by
the particular DL; on the other hand, DLs differ from their predeces-
sors, such as semantic networks and frames, in that they are equipped
with a logic-based semantics which, up to some differences in notation,
is actually the same semantics as that of classical first-order logic.
Description logics typically separate domain knowledge into two com-
ponents, a terminological part called the TBox and an assertional part
called the ABox, with the combination of a TBox and an ABox being
called a knowledge base (KB). The TBox represents knowledge about the
structure of the domain (similar to a database schema), while the ABox
represents knowledge about a concrete situation (similar to a database
1 Note that we use Description Logic (singular) to refer to the research area, and
description logics (plural) to refer to the relevant logical formalisms.

1
2 Introduction
instance). TBox statements capturing knowledge about a university do-
main might include, e.g., a teacher is a person who teaches a course,
a student is a person who attends a course and students do not teach,
while ABox statements from the same domain might include Mary is a
person, CS600 is a course and Mary teaches CS600. As already men-
tioned, a crucial feature of DLs is that such statements have a formal,
logic-based semantics. In fact the above statements can be rendered as
sentences in first-order logic as follows:

∀x (Teacher(x) ⇔ Person(x) ∧ ∃y (teaches(x, y) ∧ Course(y))),

∀x (Student(x) ⇔ Person(x) ∧ ∃y (attends(x, y) ∧ Course(y))),
∀x ((∃y teaches(x, y)) ⇒ ¬Student(x)),
Person(Mary),
Course(CS600),
teaches(Mary, CS600).
Equivalently, these statements can be written in description logic syntax
as follows:
Teacher ≡ Person  ∃teaches.Course,
Student ≡ Person  ∃attends.Course,
∃attends. ¬Student,
Mary : Person,
CS600 : Course,
(Mary, CS600) : teaches.
The first three statements of this knowledge base constitute its TBox,
and the last three its ABox. Please note how the DL syntax does not
use variables x or y. In Chapter 2 an extended version of the university
KB example will be used to define and explain DL syntax and semantics
in detail.
The logic-based semantics of DLs means that we have a well-defined,
shared understanding of when a statement is entailed by a KB; for exam-
ple, the above KB entails that Mary is a teacher. Moreover, we can use
(automated) reasoning to determine those entailments, and thus reason-
ing can be used to support the development and application of DL KBs.
Common reasoning tasks include checking the satisfiability of concepts
and the consistency of KBs, determining when one concept is more spe-
cific than another (a reasoning task called subsumption) and answering
different kinds of database-style queries over the KB.
The power of DLs derives from the fact that reasoning tasks are per-
formed with respect to the whole KB, and in particular with respect
 1.2 What are they good for and how are they used? 3
to the conceptual domain knowledge captured in the TBox. Unfortu-
nately, this power does not come without a computational cost, and
one of the most important areas of DL research has been exploring the
trade-off between the expressive power of the language available for mak-
ing statements (particularly TBox statements) and the computational
complexity of various reasoning tasks. The expressive power of DLs is
invariably constrained so as to at least ensure that common reasoning
tasks are decidable (i.e., they can always be correctly completed in a
finite amount of time), and may even be sufficiently constrained so as
to make them tractable (i.e., they can always be correctly completed in
time that is polynomial with respect to the size of the KB). In another
area of DL research, its model theory, we investigate which kinds of se-
mantic structure, i.e., interpretations or models, we can describe in a
KB. As well as theoretical investigations, e.g., determining the worst-
case complexities for various DLs and reasoning problems, there has also
been extensive practical work, e.g., developing systems and optimisation
techniques, and empirically evaluating their behaviour when applied to
benchmarks or KBs used in various applications. We will explore model
theory in Chapter 3, theoretical complexity issues in Chapter 5 and DL
reasoning techniques in Chapters 4, 6 and 7.
The emphasis on decidable and tractable formalisms is also the reason
why a great variety of extensions of basic DLs have been considered
– combining different extensions can easily lead to undecidability or
intractability, even if each of the extensions is harmless when considered
in isolation. While most DLs can be seen as decidable fragments of
first-order logic, some extensions leave the realm of classical first-order
logic, including, e.g., DLs with modal and temporal operators, fuzzy
DLs and probabilistic DLs (for details, see [BCM+ 07, Chapter 6] and
specialised survey articles such as [LWZ08, LS08]). If an application
requires more expressive power than can be provided by a decidable DL,
then one usually embeds the DL into an application program or another
KR formalism rather than using an undecidable DL.

1.2 What are they good for and how are they used?
DL systems have been used in a range of application domains, includ-
ing configuration (e.g., of telecommunications equipment) [MW98], soft-
ware information and documentation systems [DBSB91] and databases
[BCM+ 07], where they have been used to support schema design [CLN98,
BCDG01], schema and data integration [CDGL+ 98b, CDGR99], and
query answering [CDGL98a, CDGL99, HSTT00]. More recently, DLs
4 Introduction
have played a central role in the semantic web [Hor08], where they
have been adopted as the basis for ontology languages such as OWL
[HPSvH03], and its predecessors OIL and DAML+OIL, and DL know-
ledge bases are now often referred to as ontologies. This has resulted
in a more widespread use of DL systems, with applications in fields as
diverse as agriculture [SLL+ 04], astronomy [DeRP06], biology
[RB11, OSRM+ 12], defence [LAF+ 05], education [CBV+ 14], energy
management [CGH+ 13], geography [Goo05], geoscience [RP05], medicine
[CSG05, GZB06, HDG12, TNNM13], oceanography [KHJ+ 15b] and oil
and gas [SLH13, KHJ+ 15a].
In a typical application, the first step will be to determine the relevant
vocabulary of the application domain and then formalise it in a suitable
TBox. This ontology engineering process may be manual or (semi-)
automatic. In either case a DL reasoner is invariably used to check
satisfiability of concepts and consistency of the ontology as a whole. This
reasoner is often integrated in an ontology editing tool such as Protégé
[KFNM04]. Some applications use only a terminological ontology (i.e.,
a TBox), but in others the ontology is subsequently used to structure
and access data in an ABox or even in a database. In the latter case a
DL reasoner will again be used to compute query answers.
The use of DLs in applications throws the above mentioned expres-
sivity versus complexity trade-off into sharp relief. On the one hand,
using a very restricted DL might make it difficult to precisely describe
the concepts needed in the ontology and forces the modelling to remain
at a high level of abstraction; on the other hand, using a highly expres-
sive DL might make it difficult to perform relevant reasoning tasks in a
reasonable amount of time. The OWL ontology language is highly ex-
pressive, and hence also highly intractable; however, the currently used
OWL 2 version of OWL also specifies several profiles, fragments of the
language that are based on less expressive but tractable DLs. We will
discuss OWL and OWL 2 in more detail in Chapter 8.

1.3 A brief history of description logic

The study of description logic grew out of research into knowledge rep-
resentation systems, such as semantic networks and frames, and a de-
sire to provide them with precise semantics and well-defined reasoning
procedures [WS92]. Early work was mainly concerned with the im-
plementation of systems, such as Kl-One, K-Rep, Back, and Loom
[BS85, MDW91, Pel91, Mac91a]. These systems employed so-called
structural subsumption algorithms, which first normalise the concept de-
 1.3 A brief history of description logic 5
scriptions, and then recursively compare the syntactic structure of the
normalised descriptions [Neb90a]. These algorithms are usually rela-
tively efficient (polynomial), but they have the disadvantage that they
are complete only for very inexpressive DLs, i.e., for more expressive
DLs they cannot derive all relevant entailments. Early formal investi-
gations into the complexity of reasoning in DLs showed that most DLs
do not have polynomial-time inference problems [BL84, Neb90b]. Influ-
enced by these results, the implementors of the Classic system (the first
industrial-strength DL system) chose to carefully restrict the expressive
power of their DL so as to allow for tractable and complete reasoning
[PSMB+ 91, Bra92].
The so-called tableau reasoning technique for DLs was first introduced
by Schmidt-Schauß and Smolka in the early 1990s [SS91]. Tableau algo-
rithms work on propositionally closed DLs (i.e., DLs with full Boolean
operators), and are complete even for very expressive DLs. Moreover,
an implementation of one such algorithm in the Kris system showed
that, with suitable optimisations, performance on realistic problems
could be comparable with or even superior to existing structural ap-
proaches [BFH+ 92]. At the same time, there was a thorough analysis
of the complexity of reasoning in various DLs [DLNN91a, DLNN91b,
DHL+ 92], and it was observed that DLs are very closely related to modal
logics [Sch91].
Initially, tableau algorithms and systems, including Kris, considered
only relatively restricted DLs (see Section 4.2.2). On the theoretical
side, tableau algorithms were soon extended to deal with more expres-
sive DLs [HB91, Baa91, BH91, BDS93]. It took several years, however,
before the FaCT system demonstrated that suitably optimised imple-
mentations of such algorithms could be effective in practice [Hor97].
Subsequently, tableau algorithms were developed for increasingly ex-
pressive DLs [HST00], and implemented in FaCT and in other highly
optimised DL systems including Racer [HM01], FaCT++ [TH06] and
Pellet [SPC+ 07]. This line of research culminated in the development
of SROIQ [HKS06], the DL that forms the basis for the OWL ontol-
ogy language. In fact, a DL knowledge base can be seen as an OWL
ontology. The standardisation of OWL gave DLs a stable, machine-pro-
cessable and web-friendly syntax; this, and the central role of ontologies
in the semantic web, sparked an increased development of DL knowl-
edge bases (and OWL ontologies), and an increased development effort
for tools such as reasoners to determine entailments, ontology editors to
6 Introduction
write knowledge bases and APIs to programmatically access ontologies
and reasoners (see Section 8.2).
During the same period, the relationship to modal logics [DGL94a,
Sch95] and to decidable fragments of first-order logic was also studied in
more detail [Bor96, PST97, GKV97, Grä98, Grä99, LSW01], and first
applications in databases (such as schema reasoning, query optimisation,
and data integration) were investigated [LR96, BDNS98, CDGL98a,
CDGL+ 98b].
Although highly optimised implementations of tableau algorithms were
successful in many TBox reasoning applications, some larger-scale on-
tologies proved stubbornly resistant. Moreover, it remained unclear how
tableau reasoning could deal effectively with very large ABoxes. This
revived the interest in less expressive DLs, with the goal of develop-
ing tools that can deal with very large TBoxes and/or ABoxes, and
led to the development of the EL and DL-Lite families of tractable DLs
[BBL05, BBL08, CGL+ 05, CDL+ 07, ACKZ09], which are both included
in OWL 2 as profiles. A main advantage of the EL family is that it is
amenable to consequence-based reasoning techniques which scale also to
large ontologies and are more robust than tableau reasoning [BBL05].
This was first demonstrated by the CEL system [BLS06]; other relevant
implementations include ELK [KKS14] and SnoRocket [MJL13].
With the advent of the DL-Lite family of DLs, applications of de-
scription logics in databases started to receive increased interest. There
are various benefits to enriching a database application with an ontol-
ogy, such as adding domain knowledge, giving a formal definition to
the symbols used in the database and providing an enriched and unified
schema that can be used to formulate queries. These ideas have led to
the study of ontology-mediated querying [BtCLW14] and to the ontology-
based data access (OBDA) paradigm for data integration [CDL+ 09]; see
also the recent surveys [KZ14, BO15]. DL-Lite is particularly suitable
for such applications since its expressive power is sufficiently restricted
so that database-style query answering with respect to ontologies can
be reduced via query rewriting techniques to query answering in rela-
tional databases (see Chapter 7); this in turn allows standard database
systems to be used for query answering in the presence of ontologies
[CDL+ 07]. Implemented systems in this area include QuOnto and Mas-
tro [ACG+ 05, CCD+ 13] as well as Ontop [KRR+ 14].
As DLs became increasingly used, researchers investigated a multitude
of additional reasoning tasks that are intended to make DLs more usable
in various applications. These included, among many others, comput-
 1.4 How to use this book 7
ing least common subsumers and concept difference, ontology difference,
and explanation [BK06, KWW08, HPS09]. The need to support the
modularity of ontologies has been a strong driving force for studying
new reasoning problems such as module extraction [GHKS08], conser-
vative extensions [GLW06], and inseparability [BKL+ 16]. These tasks
are now widely used to support ontology engineering, and so is expla-
nation: module extraction and inseparability can be used to support
ontology reuse, e.g., by highlighting interactions between statements in
different ontologies, and explanation can be used to help debug errors
in ontologies, e.g., by highlighting the causes of inconsistencies.
Description Logic continues to be a very active research area, with
new theoretical results and new reasoning techniques and systems con-
stantly being developed; see http://dl.kr.org/. These include the
extension of tableau to hypertableau, as implemented in the HermiT
system [GHM+ 14], the extension of rewriting techniques to the EL fam-
ily of DLs and beyond [PUMH10, LTW09, BLW13, SMH13, BtCLW14],
as implemented in the KARMA [SMH13] and Grind [HLSW15] systems,
and the development of hybrid techniques, e.g., combining tableau with
consequence-based approaches in the Konclude system [SLG14].

1.4 How to use this book

This book is intended as a textbook and not as a research monograph.
Consequently, we have tried to cover all core aspects of DLs at a level
of detail suitable for a novice reader with a little background in formal
methods or logic. In particular, we expect the reader to understand the
basic notions around sets, relations and functions, e.g., their union, in-
tersection or composition. It will be useful, but not essential, for readers
to have some knowledge of first-order logic and basic notions from the-
oretical computer science. Those lacking such background may wish to
consult appropriate textbooks, e.g., http://phil.gu.se/logic/books/
Gallier:Logic_For_Computer_Science.pdf (which also contains a
nice example of a guide for readers).
This book includes both basic and advanced level material suitable
for undergraduate through to introductory graduate level courses on
description logics. In the authors’ experience, the material included here
could be covered in a 36-hour lecture course for students with a good
background in logic. For shorter courses, or those aimed at a different
cohort, some of the more advanced material can easily be dropped.
Chapters 2 and 3 provide background material, including examples
8 Introduction
and definitions, that will prove useful in the remaining chapters. Some
parts of these chapters are, however, quite long and detailed, and it may
not be appropriate to read (or teach) them in full before continuing with
the remainder of the book, but rather to dip into them as need arises.
Also, the subsequent chapters are presented in an order that the authors
find didactically convenient, but the order in which they are read and/or
taught could easily be varied.
Chapter 4 deals with tableau-based reasoning techniques; these are
typically used to reason about expressive DLs. It presents tableau algo-
rithms for ABox and KB consistency in the basic DL ALC, and shows
how they can be extended to deal with other concept and role construc-
tors. The chapter also includes a brief discussion of implementation
issues. Chapter 5 discusses the computational complexity of satisfiabil-
ity and subsumption in a variety of expressive DLs, and proves upper
and lower complexity bounds for a suitable set of these problems. It
also gives examples of extensions of DLs that are too expressive in the
sense that they lead to undecidability. Chapter 6 looks at reasoning in
the inexpressive DL EL and explains the consequence-based reasoning
technique for this logic, and it also showcases an extension (with inverse
roles) in which reasoning is more challenging. So far in this book, rea-
soning has been restricted to determining whether a DL knowledge base
entails a DL axiom. Chapter 7 discusses more complex reasoning prob-
lems, namely query answering: the entailments to be checked are from
a different language, in particular conjunctive queries and first-order
queries. Finally, Chapter 8 explains the relationship between OWL and
DLs, and describes the tools and applications of OWL.
In Chapters 2–7, citations have been kept to a minimum, but most
chapters conclude with a short section providing historical context and
a literature review.
The reader is cordially invited to actively read this book, especially
the basic definitions. Throughout the book, we provide a lot of examples
but strongly suggest that, whenever a new notion or term is introduced,
the reader should consider their own examples of this notion or term –
possibly by varying the ones presented – in order to make sure that the
newly introduced notion is completely understood. We also show how
to draw interpretations and models, and explain reasoning algorithms.
Again, in addition to the examples given, the reader should draw their
own models and run the algorithms on other inputs.
The running teaching example used throughout this book is made
available on the book’s website at http://dltextbook.org/ in an OWL
 1.4 How to use this book 9
syntax. You will also find useful further examples and exercises there,
as well as a list of errata, to which you can contribute by informing us
about any errors that you find in the book.
 2
A Basic Description Logic

In this chapter, we introduce and explain the basic notions of Descrip-

tion Logic, including syntax, semantics and reasoning services, and we
explain how the latter are used in applications.

2.1 The concept language of the DL ALC

In this section, we will describe the central notions of Description Logic
first on an intuitive level and then on a more precise level. As a running
example, we use the domain of university courses and teaching, and we
will use a conceptualisation given informally, in graphical form, in Fig-
ure 2.1. Please note that this is one way of viewing university teaching –
which might be very different from the reader’s way of viewing it. Also,
as it is an informal representation, different readers may interpret ar-
rows in different ways; that is, our representation does not come with a
well-defined semantics that would inform us in an unambiguous way how
to interpret the different arrows.1 In the next sections, we will describe
our way of viewing university teaching in a DL knowledge base, thereby
establishing some constraints on the meaning of terms like “Professor”
and “teaches” used in Figure 2.1 and throughout this section.
In Description Logic, we assume that we want to describe some ab-
straction of some domain of interest, and that this abstraction is popu-
lated by elements.2 We use three main building blocks to describe these
elements:
• Concepts represent sets of elements and can be viewed as unary pred-
1 Our graphical representation looks somewhat similar to an extended ER diagram,
for which such a well-defined semantics has been specified [Che76, CLN94].
2 We have chosen the term “elements” rather than “individuals” or “objects” to
prevent the reader from making false assumptions.

10
 2.1 The concept language of the DL ALC 11

University

offers

about
Person Course Subject
attends

Student
teaches
Teacher
UGC PGC CS Maths

TA
resp.−for
Professor

Fig. 2.1. An informal, graphical view of our running example.

icates. Concepts are built from concept names and role names (see
below) using the constructors provided by the DL used. The set a
concept represents is called its extension. For example, Person and
Course are concept names, and m is an element in the extension of
Person and c6 is in the extension of Course. To make our life a bit
easier, we often use “is a” as an abbreviation for “is in the extension
of” as, for example, in “m is a Person”.
• Role names stand for binary relations on elements and can be viewed
as binary predicates. If a role r relates one element with another
element, then we call the latter one an r-filler of the former one. For
example, if m teaches c6, then we call c6 a teaches-filler of m.

At the heart of a specific DL, we find a concept language; that is, a formal
language that allows us to build concept descriptions (and role descrip-
tions) from concept names, role names, and possibly other primitives.
For example, Person∃teaches.Course is such a concept description built
from the concept names Person and Course and the role name teaches.
Next, we formalise the exact meaning of these notions.

Definition 2.1. Let C be a set of concept names and R be a set of

role names disjoint from C. The set of ALC concept descriptions over
C and R is inductively defined as follows:

• Every concept name is an ALC concept description.

• and ⊥ are ALC concept descriptions.
• If C and D are ALC concept descriptions and r is a role name, then
the following are also ALC concept descriptions:
C  D (conjunction),
12 A Basic Description Logic
C D (disjunction),
¬C (negation),
∃r.C, (existential restriction), and
∀r.C (value restriction).

As usual, we use parentheses to clarify the structure of concepts.

Definition 2.1 fixes the syntax of ALC concept descriptions; that is,
it allows us to distinguish between expressions that are well formed and
those that are not. For example, ∃r.C and A  ∃r.∀s.(E  ¬F ) are ALC
concept descriptions, whereas ∃C and ∀s.s are not; in the former case
since ∃C is missing a role name, and in the latter case since s cannot be
both a concept and a role name.
Next, we will introduce some DL parlance and abbreviations. First,
we often use “ALC concept” as an abbreviation of “ALC concept de-
scription” and, if it is clear from the context that we talk about ALC
concepts, we may even drop the ALC and use “concepts” for “ALC
concepts”. Moreover, when clear from the context, C and R are not
mentioned explicitly.

Remark. Please note that in the DL setting a concept is, basically, a

string, and is not to be confused with the notion of a “concept” in the
sense of an abstract or general idea from philosophy. When we use a
DL in an application, we may use a DL concept to describe a relevant
application “concept”, but the latter is far more subtle and intricate
than the former.

Second, we sometimes distinguish between atomic and compound (also

called complex) concepts. An atomic concept consists of a single lexical
token, i.e., in ALC, a concept name, , or ⊥. A compound concept is
constructed using at least one of the available operators, i.e., in ALC,
, , ¬, ∃ and ∀. In the following, we will use upper case letters A,
B for concept names, upper case letters C, D for possibly compound
concepts, and lower case letters r, s for role names.
Before we define the semantics, i.e., the meaning of concepts and roles,
we will present an intuitive reading for compound concepts.

• A negation is written ¬Student and can be read as “not Student”. It

describes everything that is not in the extension of Student.
• A conjunction is written StudentTeacher and can be read as “Student
and Teacher”. It describes those elements that are in the extension of
both Student and Teacher.
 2.1 The concept language of the DL ALC 13
• A disjunction is written Student Teacher and can be read as “Student
or Teacher”. It describes those elements that are in the extension of
either Student or Teacher, or both.
• A value restriction is written ∀teaches.Course and can be read as “all
teaches-fillers are Courses”. It describes all those elements that have
only elements in Course related to them via teaches. It is written with
an upside-down A because of the “all” in its reading and its close
relationship with universal quantification in first-order logic.
• An existential restriction is written ∃teaches.Course and can be read
as “there exists a teaches-filler which is a Course”. It describes all
elements that have at least one teaches-filler that is in Course. It is
written with a backwards E because of the “there exists” in its reading
and its close relationship with existential quantification in first-order
logic.

Now, to fix the meaning of concepts and roles, we make use of an inter-

pretation, that is, a structure that:

• consists of a non-empty set called its interpretation domain. We call

the elements of this interpretation domain simply “elements”, but
they are sometimes called individuals or objects elsewhere;
• fixes, for each concept name, its extension – that is, it tells us, for each
concept name, which of the elements is (or isn’t) in the extension of
this concept;
• fixes, for each role name, its extension – that is, it tells us, for each
role name, which pairs of elements are related to each other by this
role.

Interpretations, as well as the extension of concept descriptions, are

defined next.

Definition 2.2. An interpretation I = (ΔI , ·I ) consists of a non-empty

set ΔI , called the interpretation domain, and a mapping ·I that maps

• every concept name A ∈ C to a set AI ⊆ ΔI , and

• every role name r ∈ R to a binary relation rI ⊆ ΔI × ΔI .
14 A Basic Description Logic
The mapping ·I is extended to , ⊥ and compound concepts as follows:
I
= ΔI ,
I
⊥ = ∅,
(C  D)I = C I ∩ DI ,
(C D)I = C I ∪ DI ,
(¬C)I = ΔI \ C I ,
(∃r.C)I = {d ∈ ΔI | there is an e ∈ ΔI with (d, e) ∈ rI and e ∈ C I },
(∀r.C)I = {d ∈ ΔI | for all e ∈ ΔI , if (d, e) ∈ rI , then e ∈ C I }.

We call

• C I the extension of C in I,
• b ∈ ΔI an r-filler of a in I if (a, b) ∈ rI .

Please note that an interpretation is not restricted other than as ex-

plicitly specified above: its domain must be non-empty, but can be of
any cardinality, and in particular it can be infinite; the extension of a
concept can have any number of elements between “none” and “all”;
and a role can relate any number of pairs of elements, from “none” to
“all”.
Also, please note that AI stands for the result of applying the mapping
I
· to the concept name A; this is an unusual way of writing mappings,
yet it is quite helpful and ink-saving. In the past, DL researchers have
used different notations such as I(A) or [[A]]I , but the one used here is
the one that stuck.
As an example, let us consider the following interpretation I:

ΔI = {m, c6, c7, et},

TeacherI = {m},
CourseI = {c6, c7, et},
PersonI = {m, et},
PGCI = {c7},
teaches I = {(m, c6), (m, c7), (et, et)}.

We can easily modify I to obtain other interpretations I1 , I2 etc., by

adding or removing elements and changing the interpretation of concept
and role names. An interpretation is often conveniently drawn as a di-
rected, labelled graph with a node for each element of the interpretation
domain and labelled as follows: a node is labelled with all concept names
the corresponding element of the interpretation domain belongs to, and
we find an edge from one node to another labelled with r if the element
 2.1 The concept language of the DL ALC 15

c7 Course
Course et Person
teaches PGC
m teaches teaches
Person c6
Teacher Course

Fig. 2.2. A graphical representation of the example interpretation I.

corresponding to the latter node is an r-filler of the element correspond-

ing to the former node. As an example, Figure 2.2 shows a graphical
representation of I.
Let us take a closer look at I. By definition, all elements are in the
extension of , and no element is in the extension of ⊥. The elements m
and et are, for example, in the extension of Person, and et is a teaches-
filler of itself. If we extend I to compound concepts as specified in
Definition 2.2, then we can see that, for example, m is in the extension
of PersonTeacher, and c6 is in the extension of Course¬Person, because
c6 is a Course and not a Person. Similarly, for existential restrictions,
m and et are in the extension of ∃teaches.Course, but only m is in the
extension of ∃teaches.¬Person. For value restrictions, all elements are
in the extension of ∀teaches.Course: for m and et, this is clear, and
for c6 and c7, this is because they do not have any teaches-fillers, and
hence all their teaches-fillers vacuously satisfy any condition we may
impose. In general, if an element has no r-filler, then it is in the extension
of ∀r.C for any concept C. In contrast, m is not in the extension of
∀teaches.(Course  PGC) because m has c6 as a teaches-filler that is not
in the extension of Course  PGC since it is not a PGC. At this stage, we
repeat our invitation to the reader to consider some more interpretations
and concepts and determine which element is in the extension of which
concept.
We can also investigate extensions of more compound concept descrip-
tions such as Person∃teaches.(Course¬PGC): for example, m is in the
extension of this concept since it is in the extension of both conjuncts:
by definition of I, it is in the extension of Person, and it also is in the
extension of the second conjunct, because it has a teaches-successor, c6,
that is in the extension of (Course  ¬PGC).

We have been rather generous in our syntax definition since we pro-

16 A Basic Description Logic
vide a non-minimal set of concept constructors, i.e., one with “syntactic
sugar”. The following lemma makes this observation precise.

Lemma 2.3. Let I be an interpretation, C, D concepts, and r a role.

Then
I
(i) = (C ¬C)I ,
I
(ii) ⊥ = (C  ¬C)I ,
(iii) (¬¬C)I = CI ,
(iv) ¬(C  D)I = (¬C ¬D)I ,
(v) ¬(C D)I = (¬C  ¬D)I ,
(vi) (¬(∃r.C))I = (∀r.¬C)I ,
(vii) (¬(∀r.C))I = (∃r.¬C)I .
Proof. These equations follow rather immediately from Definition 2.2:
By definition, I = ΔI , and (C ¬C)I = C I ∪ (¬C)I = C I ∪ (ΔI \
C I ) = ΔI . Hence Equation (i) holds.
Equation (ii) can be proven analogously and is left to the reader.
For Equation (iii), (¬¬C)I = (ΔI \ (ΔI \ C I )), which is of course the
same as C I .
For Equation (iv), which is also known as one of de Morgan’s laws,
we have ¬(C  D)I = ΔI \ (C  D)I . Now d ∈ ΔI \ (C  D)I if and
only if d ∈ C I or d ∈ DI (or both), which is the case if and only if
d ∈ (¬C)I ∪ (¬DI ) = (¬C ¬D)I .
Equation (v), another of de Morgan’s laws, can be proven analogously.
For Equation (vi), by definition of the semantics,
(¬(∃r.C))I
= ΔI \ {d ∈ ΔI | there is an e ∈ ΔI with (d, e) ∈ rI and e ∈ C I }
= {d ∈ ΔI | there is no e ∈ ΔI with (d, e) ∈ rI and e ∈ C I }
= {d ∈ ΔI | for all e ∈ ΔI if (d, e) ∈ rI then e ∈ C I }
= (∀r.¬C)I .
Equation (vii) can be proven analogously and is left to the reader.

As a consequence of Lemma 2.3, and as we will see later, we can

rewrite, e.g., (C D) to ¬(¬C¬D), and thus avoid explicit disjunctions.

2.2 ALC knowledge bases

If we were to use a DL-based system in an application, we would build
concept descriptions that describe relevant notions from this application
domain. For example, for a molecular biology application, we would
 2.2 ALC knowledge bases 17
build concepts describing proteins, genes and so on. We would then use
these concepts in a knowledge base, and we can do this in (at least) four
different ways:

(i) As in an encyclopedia, we define the meaning of some concept

names in terms of concept descriptions. For example, we can
define the meaning of UG-Student and CS-Teacher using the fol-
lowing equations:3
UG-Student ≡ Student  ∀attends.UGC,
CS-Teacher ≡ Teacher  ∃teaches.(Course  ∃about.CS).
Intuitively, the first equation says that UG-Students are those
students that attend only UGCs, and the second one says that
CS-Teachers are those Teachers that teach some Course about CS.
(ii) We express background knowledge. For example, we can state
that an undergraduate course (UGC) cannot be a postgraduate
course (PGC), and that a University necessarily offers both UGCs
and PGCs, using the following equations:
UGC ¬PGC,
University ∃offers.UGC  ∃offers.PGC.

(iii) We assert that individual names stand for instances of (possibly

compound) concept descriptions. For example, we can say that
Mary stands for an instance of Teacher  ∃teaches.PGC and CS600
stands for an instance of Course.
(iv) We relate individual names by roles. For example, we can say
that Mary teaches CS600.

Traditionally, we distinguish two parts of a DL knowledge base. The

terminological part, called the TBox, contains statements of the form
described in items (i) and (ii), and the assertional part, called the ABox,
contains statements of form described in items (iii) and (iv). If we
compare this to databases, then we can view a TBox as a schema because
it expresses general constraints on what (our abstraction of) the world
looks like. And we can view the ABox as the data since it talks about
concrete elements, their properties and their relationships.

2.2.1 ALC TBoxes

We start by defining the syntax and semantics of TBoxes.
3 The exact meaning of these equations will be defined later.
18 A Basic Description Logic
Definition 2.4. For C and D possibly compound ALC concepts, an
expression of the form C D is called an ALC general concept in-
clusion and abbreviated GCI . We use C ≡ D as an abbreviation for
C D, D C.
A finite set of GCIs is called an ALC TBox .
An interpretation I satisfies a GCI C D if C I ⊆ DI . An inter-
pretation that satisfies each GCI in a TBox T is called a model of T .

As usual, if it is clear that we are talking about ALC concepts and

TBoxes, we omit “ALC” and use simply TBox or GCI. We will some-
times refer to abbreviations of the form C ≡ D as an equivalence axiom,
and use axiom to refer to either an equivalence axiom or a GCI.
The interpretation I given in Figure 2.2 satisfies each of the GCIs in
T1 = {Teacher Person,
PGC ¬Person,
Teacher ∃teaches.Course,
∃teaches.Course Person}
and thus I is a model of T1 . To verify this, for each GCI C D, we de-
termine C I and DI and then check whether C I is indeed a subset of DI .
For the first GCI, we observe that TeacherI = {m} ⊆ {m, et} = PersonI .
Similarly, for the second one, we have PGCI = {c7} ⊆ {c6, c7} =
(¬Person)I . For the third one, m is the only element in the exten-
sion of Teacher and also in the extension of ∃teaches.Course, hence it is
also satisfied by I. Finally (∃teaches.Course)I = {m, et} and both m
and et are in the extension of Person.
In contrast, I does not satisfy the GCIs

Course ¬Person, (2.1)

∃teaches.Course Teacher, (2.2)

because et is both a Person and a Course, and because et teaches some

Course, but is not a Teacher.
In general, a TBox T allows us to distinguish between those inter-
pretations that are and those that are not models of T . In practice,
this means that we can use a TBox to restrict our attention to those
interpretations that fit our intuitions about the domain. For example,
Formula (2.1) should be in our TBox if we think that a Course cannot
be a Person, and Formula (2.2) should be in our TBox if we think that
only Teachers can teach Courses. In general, the more GCIs our TBox
contains, the fewer models it has. This is expressed in the following
lemma.
 2.2 ALC knowledge bases 19

Tex = {Course  ¬Person, (Tex .1)

UGC  Course, (Tex .2)
PGC  Course, (Tex .3)
Teacher ≡ Person  ∃teaches.Course, (Tex .4)
∃teaches.  Person, (Tex .5)
Student ≡ Person  ∃attends.Course, (Tex .6)
∃attends.  Person } (Tex .7)

Fig. 2.3. The example TBox Tex .

Lemma 2.5. If T ⊆ T  for two TBoxes T , T  , then each model of T 

is also a model of T .
Proof. The proof is rather straightforward: let T ⊆ T  be two TBoxes
and I a model of T  . By definition, I satisfies all GCIs in T  and thus,
since T ⊆ T  , also all GCIs in T . Hence I is, as required, also a model
of T .
Next, in Figure 2.3, we define a TBox Tex that partially captures
our intuition of teaching as presented in Figure 2.1. Axiom Tex .4 is an
equivalence that defines a Teacher as a Person who teaches a Course.
That is, every Teacher is a Person who teaches a Course and, vice versa,
if a Person teaches a Course, then they are a Teacher. Axiom Tex .5
ensures that only Persons can teach a course. As mentioned above, the
interpretation depicted in Figure 2.2 is not a model of Tex since it violates
axiom Tex .1.

2.2.2 ALC ABoxes

Next, we define ABoxes and knowledge bases.
Definition 2.6. Let I be a set of individual names disjoint from R and
C. For a, b ∈ I individual names, C a possibly compound ALC concept,
and r ∈ R a role name, an expression of the form
• a : C is called an ALC concept assertion, and
• (a, b) : r is called an ALC role assertion.
A finite set of ALC concept and role assertions is called an ALC ABox .
An interpretation function ·I is additionally required to map every
individual name a ∈ I to an element aI ∈ ΔI . An interpretation I
satisfies
• a concept assertion a : C if aI ∈ C I , and
20 A Basic Description Logic
• a role assertion (a, b) : r if (aI , bI ) ∈ rI .
An interpretation that satisfies each concept assertion and each role
assertion in an ABox A is called a model of A .
Again, if it is clear that we are talking about ALC concepts and
ABoxes, we omit “ALC” and use simply ABox, concept assertion etc.
Moreover, for the sake of brevity, we will occasionally use “individual”
as an abbreviation for “individual name”.
In Figure 2.4, we present an example ABox Aex with concept and role
assertions. The following interpretation I is a model of this ABox:
ΔI = {h, m, c6, p4},
MaryI = m,
BettyI = HugoI = h,
CS600I = c6,
Ph456I = p4,
PersonI = {h, m, c6, p4},
TeacherI = {h, m},
CourseI = {c6, p4},
PGCI = {p4},
UGCI = {c6},
StudentI = ∅,
teaches I = {(m, c6), (h, p4)},
attends I = {(h, p4), (m, p4)}.
Please observe that the individual names Hugo and Betty are interpreted
as the same element: h. This is allowed by our definition of the seman-
tics. Some logics, including many early description logics, make the so-
called Unique Name Assumption (UNA) which requires that aI = bI
in the case a = b, and would thus rule out such an interpretation.
Throughout this book, we do not make the UNA unless it is stated
to the contrary.
We can further observe that, in I, the extension of Teacher has more
elements than strictly required by Aex : nothing in Aex requires m, c6
or p4 to be in the extension of Teacher. Moreover, I interprets the
concept UGC, although this concept isn’t mentioned in Aex . Again, all
this is allowed by our definition of the semantics. Also, please note that
I is not a model of the TBox Tex given in Figure 2.3; for example,
h ∈ (Person  ∃attends.Course)I , but h ∈ Student, and thus I does not
satisfy the axiom Tex .6.
Next, we combine TBoxes and ABoxes in knowledge bases: this allows
Discovering Diverse Content Through
Random Scribd Documents
raiment, wearing on his bold front a crown, and sallying forth on a
red steed.
Some infernal Duke would appear in his proper character, quietly
seated on a griffin; another spirit of a similar rank would display the
three heads of a serpent, a man, and a cat; he would also bestride a
viper, and carry in his hand a firebrand; another of the same stamp,
would appear like a duchess, encircled with a fiery zone, and
mounted on a camel; a fourth would wear the aspect of a boy, and
amuse himself on the back of a two-headed dragon. A few spirits,
however, would be content with the simple garbs of a horse, a
leopard, a lion, an unicorn, a night-raven, a stork, a peacock, or a
dromedary; the latter animal speaking fluently the Egyptian
language. Others would assume the more complex forms of a lion or
of a dog, with a griffin’s wings attached to each of their shoulders; or
of a bull equally well gifted; or of the same animal, distinguished by
the singular appendage of a man’s face; or of a crow clothed with
human flesh; or of a hart with a fiery tail. To certain other noble
devils were assigned such shapes as those of a dragon with three
heads, one of these being human; of a wolf with a serpent’s tail,
breathing forth flames of fire; of a she wolf exhibiting the same
caudal appendage, together with a griffin’s wings, and ejecting
hideous matter from the mouth. A lion would appear either with the
head of a branded thief, or astride upon a black horse, and playing
with a viper, or adorned with the tail of a snake, and grasping in his
paws two hissing serpents. These were the varied shapes assumed by
devils of rank. To those of an inferior order were consigned upon
earth, the duty of carrying away condemned souls. These were
described as blacker than pitch: as having teeth like lions, nails on
their fingers like those of the wild boar, on their forehead horns,
through the extremities of which, poison was emitted, having wide
ears flowing with corruption, and discharging serpents from their
nostrils, and having cloven feet[54]. But this last appendage, as Sir
Thomas Brown has learnedly proved, is a mistake, which has arisen
from the devil frequently appearing to the Jews in the shape of a
rough and hairy goat, this animal being the emblem of sin-
offerings[55].
It is worthy of farther remark, says Dr. Hibbert, that the forms of
the demons described by St. Bernard, differs little from that which is
no less carefully pourtrayed by Reginald Scott, 350 years later, and,
perhaps, by the Demonologists of the present day. “In our
childhood,” says he, “our mothers’ maids have so terrified us with an
ouglie devell having hornes on his head, fier in his mouth, and a taile
in his breech, eies like a bason, fangs like a dog, clawes like a bear, a
skin like a tiger, and a voice roaring like a lion,—whereby we start
and are afraid when we heare one cry bough.”
It is still an interesting matter of speculation worth noticing—why,
after the decay of the regular systems of demonology taught in the
middle ages, the same hideous form should still be attached to the
devil? The learned Mede has remarked, “that the devil could not
appear in human shape while man was in his integrity; because he
was a spirit fallen from his first glorious perfection; and, therefore,
must appear in such a shape which might argue his imperfection and
abasement, which was the shape of a beast; otherwise, no reason can
be given, why he should not rather have appeared to Eve in the shape
of a woman than of a serpent. But since the fall of man, the case is
altered: now we know he can take upon him the shape of man. He
appears, it seems, in the shape of man’s imperfection, either for age
or deformity, as like an old man (for so the witches say); and perhaps
it is not altogether false, which is vulgarly affirmed, that the devil
appearing in human shape, has always a deformity of some uncouth
member or other, as though he could not yet take upon him human
shape entirely, for that man himself is not entirely and utterly fallen
as he is.” Grose, with considerable less seriousness, observes, that
“although the devil can partly transform himself into a variety of
shapes, he cannot change his cloven feet, which will always mark him
under every appearance.
The late Dr. Ferriar took some trouble to trace to their real source
spectral figures, which have been attributed to demoniacal visits. In
his observations on the works of Remy, the commissioner in
Lorraine, for the trial of witches, he makes the following remark:
—“My edition of this book was printed by Vincente, at Lyons, in
1595; it is entitled Dæmonolatria. The trials appear to have begun in
1583. Mr. Remy seems to have felt great anxiety to ascertain the
exact features and dress of the demons, with whom many people
supposed themselves to be familiar. Yet nothing transpired in his
examinations, which varied from the usual figures exhibited by the
gross sculptures and paintings of the middle age. They are said to be
black faced, with sunk but fiery eyes, their mouths wide and swelling
of sulphur, their hands hairy, with claws, their feet horny and
cloven.” In another part of Dr. Ferriar’s, the following account is also
given of a case which passed under his own observation:—“I had
occasion,” he observes, “to see a young married woman, whose first
indication of illness was a spectral delusion. She told me that her
apartment appeared to be suddenly filled with devils, and that her
terror impelled her to quit the house with great precipitation. When
she was brought back, she saw the whole staircase filled with
diabolical forms, and was in agonies of fear for several days. After the
first impression wore off, she heard a voice tempting her to self
destruction, and prohibiting her from all exercises of piety. Such was
the account given by her when she was sensible of the delusion, yet
unable to resist the horror of the impression. When she was newly
recovered, I had the curiosity to question her, as I have interrogated
others, respecting the forms of the demons with which she had been
claimed; but I never could obtain any other account, than that they
were very small, very much deformed, and had horns and claws like
the imps of our terrific modern romances.” To this illustration of the
general origin of the figures of demoniacal illusions, I might observe,
that, in the case of a patient suffering under delirium tremens, which
came under my notice, the devils who flitted around his bed were
described to me as exactly like the forms that he had recently seen
exhibited on the stage in the popular drama of Don Giovanni.
With the view of illustrating other accounts of apparitions, I shall
now return to the doctrine of demonology which was once taught.
Although the leading tenets of this occult science may be traced to
the Jews and early Christians, yet they were matured by our early
communication with the Moors of Spain, who were the chief
philosophers of the dark ages, and between whom and the natives of
France and Italy, a great communication subsisted. Toledo, Seville,
and Salamanca, became the greatest schools of magic. At the latter
city, prelections on the black art were, from a consistent regard to the
solemnity of the subject, delivered within the walls of a vast and
gloomy cavern. The schoolmen taught, that all knowledge might be
obtained from the assistance of the fallen angels. They were skilled in
the abstract sciences, in the knowledge of precious stones, in
alchymy, in the various languages of mankind and of the lower
animals, in the belles lettres, in moral philosophy, pneumatology,
divinity, magic, history, and prophecy. They could controul the
winds, the waters, and the influence of the stars; they could raise
earthquakes, induce diseases, or cure them, accomplish all vast
mechanical undertakings, and release souls out of purgatory. They
could influence the passions of the mind—procure the reconciliation
of friends or foes—engender mutual discord—induce mania and
melancholy—or direct the force and objects of the sexual affections.
Such was the object of demonology, as taught by its most orthodox
professors. Yet other systems of it were devised, which had their
origin in causes attending the propagation of Christianity. For it
must have been a work of much time to eradicate the universal belief,
that the Pagan deities, who had become so numerous as to fill every
part of the universe, were fabulous beings. Even many learned men
were induced to side with the popular opinion on the subject, and did
nothing more than endeavour to reconcile it with their acknowledged
systems of demonology. They taught that such heathen objects of
reverence were fallen angels in league with the prince of darkness,
who, until the appearance of our Saviour, had been allowed to range
on the earth uncontrolled, and to involve the world in spiritual
darkness and delusion. According to the various ranks which these
spirits held in the vast kingdom of Lucifer, they were suffered, in
their degraded state, to take up their abode in the air, in mountains,
in springs, or in seas. But, although the various attributes ascribed to
the Greek and Roman deities, were, by the early teachers of
Christianity, considered in the humble light of demoniacal delusions,
yet for many centuries they possessed great influence over the minds
of the vulgar. In the reign of Adrian, Evreux, in Normandy, was not
converted to the Christian faith, until the devil, who had caused the
obstinacy of the inhabitants, was finally expelled from the temple of
Diana. To this goddess, during the persecution of Dioclesian,
oblations were rendered by the inhabitants of London. In the 5th
century, the worship of her existed at Turin, and incurred the rebuke
of St. Maximus. From the ninth to the fifteenth century, several
denunciations took place of the women who, in France and Germany,
travelled over immense spaces of the earth, acknowledging Diana as
their mistress and conductor. In rebuilding St. Paul’s cathedral, in
London, remains of several of the animals used in her sacrifices were
found; for slight traces of this description of reverence, subsisted so
late as the reign of Edward the First, and of Mary. Apollo, also, in an
early period of Christianity, had some influence at Thorney, now
Westminster. About the 11th century, Venus formed the subject of a
monstrous apparition, which could only have been credited from the
influence which she was still supposed to possess. A young man had
thoughtlessly put his ring around the marble finger of her image.
This was construed by the Cyprian goddess as a plighted token of
marriage; she accordingly paid a visit to her bridegroom’s bed at
night, nor could he get rid of his bed-fellow until the spells of an
exorcist had been invoked for his relief. In the year 1536, just before
the volcanic eruption of Mount Etna, a Spanish merchant, while
travelling in Sicily, saw the apparition of Vulcan attended by twenty
of his Cyclops, as they were escaping from the effects which the over
heating of his furnace foreboded[56].
To the superstitions of Greece and Rome, we are also indebted for
those subordinate evil spirits called genii, who for many centuries
were the subject of numerous spectral illusions. A phantasm of this
kind appeared to Brutus in his tent, prophesying that he should be
again seen at Philippi. Cornelius Sylla had the first intimation of the
sudden febrile attack with which he was seized, from an apparition
who addressed him by his name; concluding, therefore, that his
death was at hand, he prepared himself for the event, which took
place the following evening. The poet Cassius Severus, a short time
before he was slain by order of Augustus, saw, during the night, a
human form of gigantic size,—his skin black, his beard squalid, and
his hair dishevelled. The phantasm was, perhaps, not unlike the evil
genius of Lord Byron’s Manfred:—
“I see a dusk and awful figure rise
Like an infernal god from out the earth;
His face wrapt in a mantle, and his form
Robed as with angry clouds; he stands between
Thyself and me—but I do fear him not.”

The emperor Julian was struck with a spectre clad in rags, yet
bearing in his hands a horn of plenty, which was covered with a linen
cloth. Thus emblematically attired, the spirit walked mournfully past
the hangings of the apostate’s tent[57].
 We may now advert to the superstitious narratives of the middle
ages, which are replete with the notices of similar marvellous
apparitions. When Bruno, the Archbishop of Wirtzburg, a short
period before his sudden death, was sailing with Henry the Third, he
descried a terrific spectre standing upon a rock which overhung the
foaming waters, by whom he was hailed in the following words:
—“Ho! Bishop, I am thy evil genius. Go whether thou choosest, thou
art and shalt be mine. I am not now sent for thee, but soon thou shalt
see me again.” To a spirit commissioned on a similar errand, the
prophetic voice may be probably referred, which was said to have
been heard by John Cameron, the Bishop of Glasgow, immediately
before his decease. He was summoned by it, says Spottiswood, “to
appear before the tribunal of Christ, there to atone for his violence
and oppressions.”
“I shall not pursue the subject of Genii much farther. The notion of
every man being attended by an evil genius, was abandoned much
earlier than the far more agreeable part of the same doctrine, which
taught that, as an antidote to this influence, each individual was also
accompanied by a benignant spirit. “The ministration of angels,” says
a writer in the Athenian oracle, “is certain, but the manner how, is
the knot to be untied.” ’Twas generally thought by the ancient
philosophers, that not only kingdoms had their tutelary guardians,
but that every person had his particular genius, or good angel, to
protect and admonish him by dreams, visions, &c. We read that
Origin, Hierome, Plato, and Empedocles, in Plutarch, were also of
this opinion; and the Jews themselves, as appears by that instance of
Peter’s deliverance out of prison. They believed it could not be Peter,
but his angel. But for the particular attendance of bad angels, we
believe it not; and we must deny it, till it finds better proof than
conjecture.”
Such were the objects of superstitious reverence, derived from the
Pantheon of Greece and Rome, the whole synod of which was
supposed to consist of demons, who were still actively bestirring
themselves to delude mankind. But in the West of Europe, a host of
other demons, far more formidable, were brought into play, who had
their origin in Celtic, Teutonic, and even Eastern fables; and as their
existence, as well as influence, was not only by the early Christians,
but even by the reformers, boldly asserted, it was long before the
rites to which they had been accustomed were totally eradicated.
Thus in Orkney, for instance, it was customary, even during the last
century, for lovers to meet within the pale of a large circle of stones,
which had been dedicated to the chief of the ancient Scandinavian
deities. Through a hole in one of the pillars, the hands of contracting
parties were joined, and the faith they plighted, was named the
promise of Odin, to violate which was infamous. But the influence of
the Dii Majores of the Edda was slight and transient, in comparison
with that of the duergar or dwarfs, who figure away in the same
mythology, and whose origin is thus recited. Odin and his brothers
killed the giant Ymor, from whose wound ran so much blood that all
the families of the earth were drowned, except one that saved himself
on board a bark. These gods then made, of the giant’s bones of his
flesh and his blood, the earth, the waters, and the heavens. But in the
body of the monster, several worms had in the course of putrefaction
been engendered, which, by order of the gods, partook of both
human shape and reason. These little beings possessed the most
delicate figures, and always dwelt in subterraneous caverns or clefts
in the rocks. They were remarkable for their riches, their activity, and
their malevolence[58]. This is the origin of our modern faries, who, at
the present day, are described as a people of small stature, gaily drest
in habiliments of green[59]. They possess material shapes, with the
means, however, of making themselves invisible. They multiply their
species; they have a relish for the same kind of food that affords
sustenance to the human race, and when, for some festal occasion,
they would regale themselves with good beef or mutton, they employ
elf arrows to bring down their victims. At the same time, they delude
the shepherds with the substitution of some vile substance, or
illusory image, possessing the same form as that of the animal they
had taken away. These spirits are much addicted to music, and when
they make their excursions, a most exquisite band of music never
fails to accompany them in their course. They are addicted to the
abstraction of the human species, in whose place they leave
substitutes for living beings, named Changelings, the unearthly
origin of whom is known by their mortal imbecility, or some wasting
disease. When a limb is touched with paralysis, a suspicion often
arises that it has been touched by these spirits, or that, instead of the
sound member, an insensible mass of matter has been substituted in
its place.
 In England, the opinions originally entertained relative to the
duergar or dwarfs, have sustained considerable modifications, from
the same attributes being assigned to them as to the Persian peris, an
imaginary race of intelligences, whose offices of benevolence were
opposed to the spightful interference of evil spirits. Whence this
confusion in proper Teutonic mythology has originated, is doubtful;
conjectures have been advanced, that it may be traced to the
intercourse the Crusaders had with the Saracens; and that from
Palestine was imported the corrupted name, derived from the peris,
of faries; for under such a title the duegar of the Edda are now
generally recognized; the malevolent character of the dwarfs being
thus sunk in the opposite qualities of the peris, the fairies. Blessing
became in England, proverbial: “Grant that the sweet fairies may
nightly put money in your shoes, and sweep your house clean.” In
more general terms, the wish denoted, “Peace be to the house[60].
Fairies, for many centuries, have been the objects of spectral
impressions. In the case of a poor woman of Scotland, Alison
Pearson, who suffered for witchcraft in the year 1586, they probably
resulted from some plethoric state of the system, which was followed
by paralysis. Yet, for these illusive images, to which the popular
superstition of the times had given rise, the poor creature was
indicted for holding communication with demons, under which light
fairies were then considered, and burnt at a stake. During her illness,
she was not unfrequently impressed with sleeping and waking
visions, in which she held an intercourse with the queen of the
Elfland and the good neighbours. Occasionally, these capricious
spirits would condescend to afford her bodily relief; at other times,
they would add to the severity of her pains. In such trances or
dreams, she would observe her cousin, Mr. William Sympsoune, of
Stirling, who had been conveyed away to the hills by the fairies, from
whom she received a salve that would cure every disease, and of
which the Archbishop of St. Andrews deigned himself to reap the
benefit. It is said in the indictment against her, that “being in Grange
Muir with some other folke, she, being sick, lay downe; and, when
alone, there came a man to her clad in green, who said to her, if she
would be faithful, he would do her good; but she being feared, cried
out; but nae bodie came to her, so she said, if he came in God’s name,
and for the gude of her soul, it was all well; but he gaed away; he
appeared another tyme like a lustie man, and many men and women
with him—at seeing him she signed herself, and pray and past with
them, and saw them making merrie with pypes, and gude cheir and
wine;—she was carried with them, and when she telled any of these
things, she was sairlie tormented by them, and the first time she gaid
with them, she gat a sair straike frae one of them, which took all the
poustie (power) of her side frae her, and left an ill-far’d mark on her
side.
“She saw the gude neighbours make their saws (salves) with panns
and fyres, and they gathered the herbs before the sun was up, and
they cam verie fearful sometimes to her, and flaire (scared) her very
sair, which made her cry, and threatened they would use her worse
than before; and at last, they tuck away the power of her haile syde
frae her, and made her lye many weeks. Sometimes they would come
and sit by her, and promise that she should never want if she would
be faithful, but if she would speak and telle of them, they would
murther her. Mr. William Sympsoune is with them who healed her,
and telt her all things;—he is a young man, not six yeares older than
herself, and he will appear to her before the court comes;—he told
her he was taken away by them; and he bid her sign herself that she
be not taken away, for the teind of them are tane to hell every
yeare[61].”
Another apparition of a similar kind may be found on the
pamphlet which was published A. D. 1696, under the patronage of
Dr. Fowler, Bishop of Glocester, relative to Ann Jefferies, “who was
fed for six months by a small sort of airy people, called fairies.” There
is every reason to suppose, that this female was either affected with
hysteria, or with that highly excited state of nervous irritability,
which, as I have shewn, gives rise to ecstatic illusions. The account of
her first fit is the only one which relates to the present subject. In the
year 1695, says her historian, “she then being nineteen years of age,
and one day knitting in an arbour in the garden, there came over the
hedges to her (as she affirmed) six persons of small stature, all
clothed in green, and which she called fairies: upon which she was so
frightened, that she fell into a kind of convulsive fit: but when we
found her in this condition, we brought her into the house, and put
her to bed, and took great care of her. As soon as she was recovered
out of the fit, she cries out, ‘they are just gone out of the window;
they are just gone out of the window. Do you not see them?’ And
thus, in the height of her sickness, she would often cry out, and that
with eagerness; which expressions we attributed to her distemper,
supposing her light-headed.” This narrative of the girl seemed highly
interesting to her superstitious neighbours, and she was induced to
relate far more wonderful stories, upon which not the least
dependance can be placed, as the sympathy she excited eventually
induced her to become a rank impostor[62].
But besides fairies, or elves, which formed the subject of many
spectral illusions, a domestic spirit deserves to be mentioned, who
was once held in no small degree of reverence. In most northern
countries of Europe there were few families that were without a
shrewd and knavish sprite, who, in return for the attention or neglect
which he experienced, was known to
——“sometimes labour in the quern,
And bootless make the breathless housewife churn;
And sometimes make the drink to bear no barm!”

Mr. Douce, in his Illustrations of Shakspeare, has shewn, that the

Samogitæ, a people formerly inhabiting the shores of the Baltic, who
remained idolatrous so late as the 15th century, had a deity named
Putseet, whom they invoked to live with them, by placing in the barn,
every night, a table covered with bread, butter, cheese, and ale. If
these were taken away, good fortune was to be expected; but if they
were left, nothing but bad luck. This spirit is the same as the goblin-
groom, Puck, or Robin Good-fellow of the English, whose face and
hands were either of a russet or green colour, who was attired in a
suit of leather, and armed with a flail. For a much lesser fee than was
originally given him, he would assist in threshing, churning, grinding
malt or mustard, and sweeping the house at midnight[63]. A similar
tall “lubbar fiend,” habited in a brown garb, was known in Scotland.
Upon the condition of a little wort being laid by for him, or the
occasional sprinkling, upon a sacrificial stone, of a small quantity of
milk, he would ensure the success of many domestic operations.
According to Olaus Magnus, the northern nations regarded domestic
spirits of this description, as the souls of men who had given
themselves up during life to illicit pleasures, and were doomed, as a
punishment, to wander about the earth, for a certain time, in the
peculiar shape which they assumed, and to be bound to mortals in a
sort of servitude. It is natural, therefore, to expect, that these familiar
spirits would be the subjects of many apparitions, of which a few
relations are given in Martin’s Account of the Second Sight in
Scotland. “A spirit,” says this writer, “called Browny, was frequently
seen in all the most considerable families in the isles and the north of
Scotland, in the shape of a tall man; but within these twenty or thirty
years, he is seen but rarely.”
It is useless to pursue this subject much farther: in the course of a
few centuries, the realms of superstition were increased to almost an
immeasurable extent; the consequence was, that the air, the rocks,
the seas, the rivers, nay, every lake, pool, brook, or spring, were so
filled with spirits, both good and evil, that of each province it might
be said, in the words of the Roman satirist, “Nosiba regio tam plena
est numinibus, ut facilius possis deum quam hominem invenire.”
Hence the modification which took place of systems of demonology,
so as to admit of the classification of all descriptions of devils,
whether Teutonic, Celtic, or Eastern systems of mythology. “Our
schoolmen and other divines,” says Burton in his Anatomy of
Melancholy, “make nine kinds of bad devils, as Dionysius hath of
angels. In the first rank, are those false gods of the Gentiles, which
were adored heretofore in several idols, and gave oracles at Delphos
and elsewhere, whose prince is Beelzebub. The second rank is of liars
and equivocators, as Apollo, Pythias, and the like. The third are those
vessels of anger, inventors of all mischief, as that of Theutus in Plato.
Esay calls them vessels of fury: their prince is Belial. The fourth are
malicious, revengeful devils, and their prince is Asmodeus. The fifth
kind are coseners, such as belong to magicians and witches; their
prince is Satan. The sixth are those aërial devils that corrupt the air,
and cause plagues, thunders, fires, &c. spoken of in Apocalypse and
Paule; the Ephesians name them the prince of the air: Meresin is
their prince. The seventh is a destroyer, captaine of the furies,
causing wars, tumults, combustions, uproares, mentioned in the
Apocalypse, and called Abaddon. The eighth is that accusing or
calumniating devil, whom the Greeks call Διάβολος, that drives us to
despair. The ninth are those tempters in several kindes, and their
prince is Mammon.”
But this arrangement was not comprehensive enough; for, as
Burton adds, “no place was void, but all full of spirits, devils, or other
inhabitants; not so much as an haire-breadth was empty in heaven,
earth, or waters, above or under the earth; the earth was not so full of
flies in summer as it was at all times of invisible devils.”
Pneumatologists, therefore, made two grand distinctions of demons;
there were celestial demons, who inhabited the regions higher than
the moon; while those of an inferior rank, as the Manes or Lemures,
were either nearer the earth, or grovelled on the ground. Psellus,
however, “a great observer of the nature of devils,” seems to have
thought, that such a classification destroyed all distinction between
good and evil spirits: he, therefore, denied that the latter ever
ascended the regions above the moon, and contending for this
principle, founded a system of demonology, which had for its basis
the natural history and habitations of all demons. He named his first
class fiery devils. They wandered in the region near the moon, but
were restrained from entering into that luminary; they displayed
their power in blazing stars, in fire-drakes, in counterfeit suns and
moons, and in the euerpo santo, or meteoric lights, which, in vessels
at sea, flit from mast to mast, and forebode foul weather. It was
supposed that these demons occasionally resided in the furnaces of
Hecla, Etna, or Vesuvius. The second class consisted of aërial devils.
They inhabited the atmosphere, causing tempests, thunder and
lightning; rending asunder oaks, firing steeples and houses, smiting
men and beasts, showering down from the skies, stones[64], wool, and
even frogs; counterfeiting in the clouds the battles of armies, raising
whirlwinds, fires, and corrupting the air, so as to induce plagues. The
third class was terrestrial devils, such as lares, genii, fawns, satyrs,
wood-nymphs, foliots, Robin good-fellows, or trulli. The fourth class
were aqueous devils; as the various description of water-nymph, or
mermen, or of merwomen. The fifth were subterranean devils, better
known by the name dæmones itallici, metal-men, Getuli or Cobals.
They preserved treasure in the earth, and prevented it from being
suddenly revealed; they were also the cause of horrible earthquakes.
Psellus’s sixth class of devils were named lucifugi; they delighted in
darkness; they entered into the bowels of men, and tormented those
whom they possessed with phrenzy and the falling sickness. By this
power they were distinguished from earthly and aërial devils; they
could only enter into the human mind, which they either deceived or
provoked with unlawful affections.
 Nor were speculations wanting with regard to the common nature
of these demons. Psellus conceived that their bodies did not consist
merely of one element, although he was far from denying that this
might have been the case before the fall of Lucifer. It was his opinion,
that devils possessed corporeal frames capable of sensation; that
they could both feel and be felt; they could injure and be hurt; that
they lamented when they were beaten, and that if struck into the fire,
they even left behind them ashes,—a fact which was demonstrated in
a very satisfactory experiment made by some philosophers upon the
borders of Italy; that they were nourished with food peculiar to
themselves, not receiving the aliment through the gullet, but
absorbing it from the exterior surface of their bodies, after the
manner of a sponge; that they did not hurt cattle from malevolence,
but from mere love of the natural and temperate heat and moisture
of these animals; that they disliked the heat of the sun, because it
dried too fast; and, lastly, that they attained a great age. Thus,
Cardan had a fiend bound to him twenty-eight years, who was forty-
two years old, and yet considered very young. He was informed, from
this very authentic source of intelligence, that devils lived from two
to three hundred years, and that their souls died with their bodies.
The very philosophical statement was, nevertheless, combated by
other observers. “Manie,” says Scot, “affirmed that spirits were of
aier, because they had been cut in sunder and closed presentlie
againe, and also because they vanished away so suddenlie.”
“The Narrative of the Demon of Tedworth,
or the disturbances at Mr. Monpesson’s
house, caused by Witchcraft and Villainy
of the Drummer.”
“In winter’s tedious nights, sit by the fire
With good old folks; and let them tell the tales
Of woeful ages long ago betid.”

“Mr. John Monpesson of Tedworth, in the County of Wilts, being

about the middle of March, in the year 1661, at a neighbouring town
called Ludlow, and hearing a drummer beat there, he enquired of the
bailiff of the town at whose house he then was, what it meant. The
bailiff told him, that they had for some days past been annoyed by an
idle drummer, who demanded money of the constable by virtue of a
pretended pass, which he thought was counterfeited. On hearing
this, Mr. Monpesson sent for the fellow, and asked him by what
authority he went up and down the country in that manner with his
drum. The drummer answered, that he had good authority, and
produced his pass, with a warrant under the hands of Sir William
Cawley, and Colonel Ayliff, of Gretenham. Mr. Monpesson, however,
being acquainted with the hand-writing of these gentlemen,
discovered that the pass and warrant were counterfeit, upon which
he commanded the vagrant to lay down his drum, and at the same
time gave him in charge to a constable, to carry him before the next
justice of the peace, to be farther examined and punished. The fellow
then confessed that the pass and warrant were forged, and begged
earnestly to be forgiven and to have his drum restored: upon this Mr.
Monpesson told him, that if, upon enquiry from Colonel Ayliff,
whose drummer he represented himself to be, he should turn out to
be an honest man, he should listen to his entreaty and have the drum
back again; but that, in the mean time, he would take care of it. The
drum, therefore, was left in the bailiff’s hand; and the drummer went
off in charge of the constable, who, it appears, was prevailed upon, by
the fellow’s entreaties, to allow him to escape.
 About the middle of April following, at a time when Mr.
Monpesson was preparing for a journey to London, the bailiff sent
the drum to his house. On his return from his journey, his wife
informed him that they had been very much alarmed in the night by
thieves, and that the house had like to have been torn down. In
confirmation of this alarm, Mr. Monpesson had not been above three
nights at home, when the same noise was again heard which had
disturbed the family in his absence. It consisted of a tremendous
knocking at the doors, and thumping on the walls of the house; upon
which Mr. M. got out of bed, armed himself with a brace of pistols,
opened the street door to ascertain the cause, which he had no
sooner done, than the noise removed to another door, which he also
opened, went out, and walked round the house; but could discover
nothing, although he heard a strange noise and hollow sound. He
had no sooner returned and got into bed, than he was again
disturbed by a noise and drumming on the top of the house, which
continued for a length of time, and then gradually subsided, as if it
went off into the air.
The noise of thumping and drumming, after this, was very
frequent; usually for five nights together, when there would be an
intermission of three. The noise was on the outside of the house,
which principally consisted of board; and usually came on just as the
family was going to bed, whether that happened early or late. After
continuing these annoyances for a month on the outside of the
house, it at length made bold to come into the room where the drum
lay, four or five nights in every seven; coming always on after they
had got into bed, and continuing for two hours after. The signal for
the appearance of the noise was the hearing of a hurling of the air
over the house; and when it was about to retire, the drum would beat
the same as if a guard were being relieved. It continued in this room
for the space of two months, during which time Mr. Monpesson lay
there to observe it. In the early part of the night, it used to be very
troublesome, but after it had continued two hours, all would be quiet
again.
During the prevalence of this disturbance, Mrs. Monpesson was
brought to bed, and the night on which this occurrence took place,
there was but very little noise made, nor any at all for the three
subsequent weeks of her confinement. After this polite and well-
timed cessation, it returned in a sudden and more violent manner
than before; it followed and teased their youngest children, and beat
against their bedstead so violently that every moment they were
expected to be broken to pieces. On placing their hands upon them at
this time, no blows were felt, although they were perceived to shake
exceedingly. For an hour together the drum would beat roundheads
and cuckold, the tat-too, and several other martial pieces, as well as
any drummer could possibly execute them. After this, a scratching
would be heard under the children’s beds, as if something that had
iron claws were at work. It would lift the children up in their beds,
follow them from one room to another, and for a while only haunted
them, without playing any other pranks.
There was a cockloft in the house, which had not been observed to
be troubled; and to this place the children were removed; and were
always put to bed before daylight disappeared, but here they were no
sooner laid, than their disturber was at his work again with them.
On the fifth of November, 1661, a terrible noise was kept up; and
one of Mr. Monpesson’s servants observing two boards moving in the
children’s room, asked that one might be given to him; upon which a
board came (nothing moving it that he saw) within a yard of him; the
man said again, let me have it in my hand; when it was brought quite
close to him, and in this manner it was continued moving up and
down, to and fro, for at least twenty minutes together. Mr.
Monpesson, however, forbade his servant to take liberties with the
invisible and troublesome guest in future. This circumstance took
place in the day-time, and was witnessed by a whole room full of
people. The morning this occurred, it left a very offensive
sulphureous smell behind it. At night, the minister of the parish, one
Mr. Cragg, and several of the neighbours, paid Mr. M. a visit. The
minister prayed at the children’s bedside, when the demon was then
extremely troublesome and boisterous. During time of prayer it
retired into the cockloft, but as soon as prayers were over it returned;
when in the presence and sight of the company, the chairs began to
walk and strut about the room of their own accord, the children’s
shoes were thrown over their heads, and every thing loose moved
about the room. At the same time, a bedpost was thrown at the
minister, which struck him on the leg, but so gently that a lock of
wool could not have fallen more gently; and it was observed, that it
stopped just where it fell, without rolling or otherwise moving from
the place.
In consequence of the demon tormenting the children so
incessantly, he had them removed to a neighbour’s house, taking his
eldest daughter, who was about ten years of age, into his own
chamber, where it had not been for a month before; but, as soon as
she was in bed, the noise began there again, and the drumming
continued for three weeks with other noises; and if any particular
thing was called for to be beaten on the drum, it would perform it.
The children were brought home again, in consequence of the house
where they were placed being crowded with strangers. They were
now placed in the parlour, which, it was remarked, had hitherto not
been disturbed; but no sooner were they here, than their tormentor,
while they were in bed, amused himself with pulling their hair and
bedgowns, without offering any other violence.
It was remarked, that when the noise was loudest, and when it
came with the most sudden and surprising violence, no dog about
the house would move or bark, though the knocking and thumping
were often so boisterous and rude, that they were heard at a
considerable distance in the fields, and awakened the neighbours in
the village, some of whom lived very near this house. Not
unfrequently the servants would be lifted up, with their bed, to a
considerable height, and then let gently down again without harm; at
other times it would lie like a great weight upon their feet.
About the end of December, 1661, the drumming was less frequent,
but then a noise like the chinking of money was substituted for it,
occasioned, as it was thought, in consequence of something Mr.
Monpesson’s mother had said the day before to a neighbour, who
spoke about fairies leaving money behind them; viz. that she should
like it well, if it would leave them some to make them amends for the
trouble it had caused them. The following night, a great chinking and
jingling of money was heard all over the house. After this it left off its
ruder pranks, and amused itself in little apish and less troublesome
tricks. On Christmas morning, a little before daylight, one of the little
boys was hit, as he was getting out of bed, upon a sore place on his
heel, with the latch of the door, the pin of which, that fastened it to
the door, was so small, that it was a matter of no little difficulty for
any one else to pick it. The night after Christmas, it threw the old
gentlewoman’s clothes about the room, and hid her bible in the
ashes; with a number of other mischievous tricks of the same kind.
After this, it became very troublesome to one of Mr. Monpesson’s
servant men, a stout fellow, and of sober conversation. This man
slept in the house during the greater part of the disturbance; and for
several nights something would attempt to pull the bedclothes off
him, which he often, though not always, prevented by main force; his
shoes were frequently thrown at his head, and sometimes he would
find himself forcibly held, as it were, hand and feet; but he found that
when he could use a sword which he had by him, and struck with it,
the spirit let go his hold.
Some short time after these contests, a son of Mr. Thomas Bennet,
for whom the drummer had sometimes worked, came to the house,
and mentioned some words to Mr. Monpesson that the drummer
had spoken, which it seems were not well taken; for they were no
sooner in bed, than the drum began to beat in a most violent
manner: the gentleman got up and called his man, who was lying
with Mr. Monpesson’s servant just mentioned, whose name was
John. As soon as Mr. Bennet’s man was gone, John heard a rustling
noise in his chamber, as if a person in silks were moving up and
down; he immediately put out his hand for his sword, which he felt
was withheld by some one, and it was with difficulty and much
tugging, that he got it again into his possession, which he had no
sooner done, than the spectre left him; and it was always remarked it
avoided a sword. About the beginning of January, 1662, they used to
hear a singing in the chimney before it descended; and one night,
about this time, lights were seen in the house. One of them came into
Mr. Monpesson’s chamber, which appeared blue and glimmering,
and caused a great stiffness in the eyes of those who beheld it. After
the light disappeared, something was heard walking or creeping up
stairs, as if without shoes. The light was seen four or five times in the
children’s chamber; and the maids confidently affirm, that the doors
were at least ten times opened or shut in their presence; and that,
when they were opened, they heard a noise as if half a dozen had
entered together; some of which were afterwards heard to walk about
the room, and one rustled about as if it had been dressed in silk,
similar to that Mr. Monpesson himself heard.
 While the demon was in one of his knocking moods, and at a time
when many were present, a gentleman of the company said, “Satan,
if the drummer set thee to work, give three knocks and no more;”
which it did very distinctly, and stopped. The same gentleman then
knocked to hear if it would answer him as it was accustomed to do.
For further proof, he required it, if it actually were the drummer that
employed him as the agent of his malice, to give five knocks and no
more that night; which it did, and quietly left the house for the
remainder of the night. This was done in the presence of Sir Thomas
Chamberlaine of Oxfordshire, and many other creditable persons.
On Saturday morning, an hour before daylight, January 10, a drum
was heard beating upon the outside of Mr. Monpesson’s chamber,
from whence it went to the other end of the house, where some
gentlemen strangers lay, and commenced playing at their door four
or five different tunes; and at length flew off in the air. The next
night, a blacksmith in the village, and Mr. Monpesson’s man John,
who was lying with him, heard a noise in the room, as if somebody
were shoeing a horse; and something came with something like a
pair of pincers, and nipped at the blacksmith’s nose the whole of the
night.
Getting up one morning to go a journey, Mr. Monpesson heard a
great noise below, where the children lay; and on running down
instantly with a pistol in his hand, he heard a voice cry out, a witch! a
witch! similar to one they had heard on a former occasion. On his
entering the apartment, all became quiet again.
The demon having one night played some little pranks at the foot
of Mr. Monpesson’s bed, it went into another bed, where one of his
daughters lay, and passed from one side to the other, lifting her up as
it passed under her. At that time there were three kinds of noises in
the bed. They attempted to thrust at it with a sword, but it
continually evaded them. The following night it came panting like a
dog out of breath, when some one present took a bedpost to strike at
it, when it was immediately snatched out of her hand; and company
coming up stairs at the same time, the room was filled with a
nauseous stench, and very hot, although there was no fire on, and
during a very sharp winter’s night. It continued panting an hour and
a half, panting and scratching; and afterwards went into the
adjoining chamber, where it began to knock a little, and seemed to
rattle a chair; thus it continued for two or three nights in succession.
The old lady’s bible after this was found again among the ashes, with
the leaves downwards. It was taken up by Mr. Monpesson, who
observed that it lay open at the third chapter of St. Mark, where
mention is made of the unclean spirits falling down before our
Saviour, and of his giving power to the twelve Apostles to cast out
devils, and of the Scribes’ opinion, and that he cast them out through
Beelzebub.
The following morning ashes were scattered over the chamber
floor, to see what impressions would be left upon it; in the morning,
in one place they found the resemblance of a great claw in another
that of a smaller one, some letters in another, which could not be
decyphered, besides a number of circles and scratches in the ashes,
which no one understood except the demon itself.
About this time, the author of the narration went to the house to
enquire after the truth of the circumstances which made so much
noise in that part of the country. The demon had left off drumming,
and the terrible noises it was in the habit of making before he
arrived; but most of the remarkable facts already related, were
confirmed to him there by several of the neighbours, on whose
veracity he could depend, who had witnessed them. It now used to
haunt the children after they were gone to bed. On the night he was
there, the children went to bed about 8 o’clock; a maid servant
immediately came down and informed us that the spirit was come.
The neighbours then present went away, as well as two ministers
who had previously been some time in the house, but Mr.
Monpesson the author, and another gentleman who came with him,
went up to the room where the children were in bed. A scratching
was heard as they went up stairs, and just as they got into the room,
it was perceived just behind the bolster of the bed in which the
children lay, and appeared to be lying against the tick. The noise it
made was like that made with long nails upon the bolster. There were
two little girls, about seven or eight years of age, in the bed. Their
hands were outside the bedclothes, so that it was perfectly visible the
noise was not made by them which was behind their heads: they had
been so used to it of late, and always with some present in the
chamber, that they seemed to take very little notice of it. The
narrator, who was standing at the head of the bed, thrust his hand
Welcome to our website – the ideal destination for book lovers and
knowledge seekers. With a mission to inspire endlessly, we offer a
vast collection of books, ranging from classic literary works to
specialized publications, self-development books, and children's
literature. Each book is a new journey of discovery, expanding
knowledge and enriching the soul of the reade

Our website is not just a platform for buying books, but a bridge
connecting readers to the timeless values of culture and wisdom. With
an elegant, user-friendly interface and an intelligent search system,
we are committed to providing a quick and convenient shopping
experience. Additionally, our special promotions and home delivery
services ensure that you save time and fully enjoy the joy of reading.

Let us accompany you on the journey of exploring knowledge and

personal growth!

textbookfull.com