Beruflich Dokumente
Kultur Dokumente
Textbook New Methods To Engineer and Seamlessly Reconfigure Time Triggered Ethernet Based Systems During Runtime Based On The Profinet Irt Example 1St Edition Lukasz Wisniewski Ebook All Chapter PDF
Textbook New Methods To Engineer and Seamlessly Reconfigure Time Triggered Ethernet Based Systems During Runtime Based On The Profinet Irt Example 1St Edition Lukasz Wisniewski Ebook All Chapter PDF
https://textbookfull.com/product/back-trouble-a-new-approach-to-
prevention-and-recovery-based-on-the-alexander-technique-deborah-
caplan/
https://textbookfull.com/product/practical-system-programming-
with-c-pragmatic-example-applications-in-linux-and-unix-based-
operating-systems-1st-edition-sri-manikanta-palakollu/
https://textbookfull.com/product/practical-system-programming-
with-c-pragmatic-example-applications-in-linux-and-unix-based-
operating-systems-1st-edition-sri-manikanta-palakollu-2/
https://textbookfull.com/product/multivariate-techniques-an-
example-based-approach-1st-edition-saman-hanif-shahbaz-muhammad-
hanif/
Biometric Identification Technologies Based on Modern
Data Mining Methods Stepan Bilan
https://textbookfull.com/product/biometric-identification-
technologies-based-on-modern-data-mining-methods-stepan-bilan/
https://textbookfull.com/product/cases-on-research-based-
teaching-methods-in-science-education-1st-edition-eugene-de-
silva/
https://textbookfull.com/product/location-based-social-media-
space-time-and-identity-1st-edition-leighton-evans/
https://textbookfull.com/product/mastering-kafka-streams-and-
ksqldb-building-real-time-data-systems-by-example-1st-edition-
mitch-seymour/
https://textbookfull.com/product/wearable-computing-from-
modeling-to-implementation-of-wearable-systems-based-on-body-
sensor-networks-1st-edition-giancarlo-fortino/
Technologien für die intelligente Automation
Technologies for Intelligent Automation
Lukasz Wisniewski
Herausgegeben von
inIT – Institut für industrielle Informationstechnik
Hochschule Ostwestfalen-Lippe
Lemgo, Germany
Lukasz Wisniewski
Springer Vieweg
© Springer-Verlag GmbH Germany 2017
This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part of the
material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation,
broadcasting, reproduction on microfilms or in any other physical way, and transmission or information storage
and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology now known or
hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication does
not imply, even in the absence of a specific statement, that such names are exempt from the relevant protective
laws and regulations and therefore free for general use.
The publisher, the authors and the editors are safe to assume that the advice and information in this book are
believed to be true and accurate at the date of publication. Neither the publisher nor the authors or the editors
give a warranty, express or implied, with respect to the material contained herein or for any errors or omissions
that may have been made. The publisher remains neutral with regard to jurisdictional claims in published maps
and institutional affiliations.
I would like to thank all the people who in some way contributed and helped me to
finish this thesis.
First of all, I would like to thank Professor Christian Diedrich, for accepting me into
his group and giving me a freedom during my Ph.D research. I’m very thankful for his
guidance and critical feedback to my work. It was really an honour to work under his
supervision.
My special and heartily thanks to Professor Jürgen Jasperneite for giving me the op-
portunity to join the Institute Industrial IT and pursuit my work towards Ph.D. I would
like to thank him for his substantial as well as mental support during the whole du-
ration of my thesis. I’m grateful for the superb environment and the opportunity to
attend many conferences around the world, where I could share and discuss results of
my research.
I would like to also thank Professor Thilo Sauter who kindly agreed to serve as third
reviewer of my thesis.
I would like to especially thank to my colleagues and also ex-colleagues of the Insti-
tute Industrial IT and Fraunhofer IOSB-INA in Lemgo for their countless helpful discus-
sions and many inspirations that helped me in completing this work. Moreover, many
thanks to Henning Trsek, Shiraz Gilani, Tim Tack, Uwe Mönks, Verena Wendt, Arne Neu-
mann, Dominik Henneke, Lars Dürkop, Sergej Gamper, Carsten Pieper, Alexander Bien-
darra, Sebastian Schriegel and Holger Flatt for all discussions and the help in improving
the quality of this work. Special thanks goes to Markus Schumacher for his continuous
support and help in validation of many ideas. Your support was greatly appreciated.
Lastly, I would like to thank my family for all their support and encouragement during
this exciting but also sometimes difficult time. I want to thank my wife Gosia for taking
care of everything, especially our two sons Alan and Ian and allowing me to focus on my
Ph.D work. I greatly appreciate your faithful support and patience, especially during the
final phase of this thesis. Thank you!
Contents
1 Introduction 1
1.1 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Objective . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.3 Outline of the document . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2 Fundamentals 7
2.1 Production Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.1.1 Automation Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.1.2 Networked Control System . . . . . . . . . . . . . . . . . . . . . . . . 9
2.2 NCS Communication Requirements . . . . . . . . . . . . . . . . . . . . . . . 11
2.2.1 Timeliness in NCS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
Data Collection and Data Distribution . . . . . . . . . . . . . . . . . 13
NCS System Reaction Time . . . . . . . . . . . . . . . . . . . . . . . . 15
Deterministic Communication . . . . . . . . . . . . . . . . . . . . . . 16
2.2.2 Communication Reliability in NCS . . . . . . . . . . . . . . . . . . . . 19
2.2.3 Industrial Communication Requirements - Summary . . . . . . . . 21
2.3 Industrial Network Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
2.3.1 Industrial Ethernet based Protocols - Real-Time . . . . . . . . . . . . 23
Class 3 – Polling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
Class 3 – Token Passing . . . . . . . . . . . . . . . . . . . . . . . . . . 24
Class 3 – TDMA Based . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
2.3.2 Industrial Ethernet based Protocols - Reliability . . . . . . . . . . . . 25
Physical Layer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
Data Link Layer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
Application Layer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
2.4 Engineering of Industrial Network Systems . . . . . . . . . . . . . . . . . . . 28
2.5 Re-engineering of Industrial Network Systems . . . . . . . . . . . . . . . . . 30
2.6 Problem Definition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
2.7 Solution Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
Bibliography 175
Abbreviations
ADAG Almost Directed Acyclic Graph
AP Automation Pyramid
AS Automation System
BE Best Effort
CC Communication Cycle
CM Compression Master
CN Controlled Node
CS Control System
DC Distributed Clock
GM Grand Master
IOD IO Device
IT Information Technology
LP Linear Program
MN Managing Node
P2P Peer-to-Peer
PM Production Module
RC Rate-Constraint
RT Real-Time
SC Synchronisation Client
SM Synchronisation Master
SR Stream Reservation
TC Transparent Clock
TP Technical Process
TT Time Triggered
The current technical processes in the industry are controlled by the Programmable
Logic Controllers (PLCs), which connected together with Sensors and Actuators via net-
work system form a Networked Control System (NCS)[1]. Depending on the type of
technical process, requirements in terms of sampling rates of controlled processes and
corresponding reaction times may differ [2]. They may require reaction times in the
range of few milliseconds, in case of motion control applications, up to hundreds of
milliseconds or more in case of process automation applications, such as control of
biochemical processes. Although, some specific requirements for the network system
responsible to transit data between PLC and Sensor and Actuator may vary, all of them
require to exchange process data in real-time. Real-time communication is achieved
through introduction of special prioritisation rules and dedicated hardware. In order to
achieve determinism and cycle times that are equal or less than 1ms, the time critical
communication has to be specially organised and isolated from other kinds of traffic.
Such highly deterministic network systems have one additional thing in common that
is limited flexibility. When it comes to changes of such systems, typically an offline
engineering process in a special tool has to be performed. Afterwards, the controlled
process has to be stopped and the communication system is restarted.
In this dissertation a concept is proposed that allows to implement selected engi-
neering functionalities into a field level devices, such as PLC. Thus allowing to perform
reconfiguration of industrial networks systems, which are part of NCSs in a seamless
way.
1.1 Motivation
Several industrial network systems (INSs) supporting exchange of information in real-
time had been developed in the past years. They can be categorised into three main
real-time performance classes, as defined in [3]. The first class supports low real-time
performance, i.e. cycle times around 100ms, which can be achieved with the standard-
ised network equipment. The second class supports medium performance, i.e. cycle
times around 10ms. This performance is typically realised by prioritised real-time traf-
fic delivery, using slightly modified network hardware. In order to achieve the high-
est real-time performance with the cycle times lower than 1ms, independent on the
amount of traffic present in the network system, special communication rules have
to be created (e.g. scheduled communication) and sophisticated hardware is needed.
This performance is achieved by Time Triggered Ethernet based (TTEB) protocols, rep-
resented by technologies, such as PROFINET IRT[4][5], EtherCAT [6][7][8], TTEthernet
[9], Powerlink[10], SERCOS III [11] or the newest IEEE 802.1 standard, which is still un-
der development, Times Sensitive Networking (TSN) [12][13][14][15][16]. All of them
are capable of providing high real-time performance, with cycle times going down to
12,5μ as in case of EtherCAT [17]. Apart from the real-time performance, these systems
also provide high level of reliability, which is required in case of critical applications.
This is typically realised by diverse redundancy approaches [18].
The current trends, such as Internet of Things (IoT) [19], vertical integration of the au-
tomation system using technologies such as OPC UA [20], camera based surveillance
of technical processes or camera based localisation [21] will be a big challenge for the
INSs. This is due to the increased amount of traffic produced by the additional net-
work members and their potential influence on the real-time communication. Here,
an important role can be played by TTEB systems. With the help of precisely defined
communication rules, such as Time Division Multiple Access (TDMA) with commu-
nication scheduling, token passing or polling based approaches [22], they allow the
assignment of critical data to dedicated, exclusive time slots, providing deterministic
data exchange. Such communication systems can be used not only to service the most
demanding application classes as the motion control or safety critical applications, but
also give the possibility to deal with large systems, such as automotive production sys-
tems, consisting of thousands of network entities. The possibility to assign a fraction of
bandwidth to selected traffic for providing defined, constant performance at any given
time, independent to the number of users and the current traffic load in the system,
may become a key feature in a crowded network systems.
In order to remain competitive, the future production systems have to react promptly
on the market turbulences, to satisfy diverse customer needs. Currently, flexibility of
production systems gives some angle of possible changes but for a hefty price. The
price is typically related to the retrofitting of production line, which is time consuming
and expensive due to necessity of manual engineering and configuration of machines,
robots, network devices, etc., and results in production stops. As a consequence, there
must always be a minimum amount of workpieces (the minimum number typically de-
pends on the product complexity) that have to be produced to reach the cost efficient
level. The current trend of the production paradigm is the mass customisation, which
suppose to allow the customer to define features of the ordered product (even at the
level of the computer-aided design (CAD)). For the production system, it would mean
that there might be a need for a production with the lot size 1. Therefore, the pro-
duction system should be capable of handling such requirements. The paradigm that
allows customisation of products, while keeping production costs at a reasonable level,
is the agile manufacturing [23], [24], [25]. Figure 1.1 illustrates the relationship between
a customer and an agile manufacturing system as a use case diagram. The possible
interactions can be defined as:
• Customise product – allows to exactly specify the product that the customer is
ordering. The customer is even able to change the shape of the product, e.g. in
order to easily integrate it into some machine.
1.1. Motivation 3
• Order product – allows to order even one piece. Modular structure of the pro-
duction lines together with production scheduling software allow to achieve cost
efficient level even in case of lot size equal to 1.
• Deliver product – there are several ways to perform delivery, but it is typically
done by shipment.
In order to allow realisation of such systems, there are several aspects that should be
tackled at different levels of the production system [26]. Starting from the highest levels,
such as:
• collection and visualisation of plant data of Supervisory Control and Data Acqui-
sition (SCADA) tools at the cell level,
• issues related to engineering of PLCs, sensors and actuators at the field level.
The main focus of this dissertation is the lowest level, where the NCSs are located.
into the family of hard mathematical problems [29][30]. This might be also the reasons
why, currently, only a very few companies offer a complete NCSs based on TTEB, such
as PROFINET IRT by Siemens, including PLCs, IODs and an engineering tool. Addi-
tionally, it has to be mentioned that none of the currently available solutions supports
flexibility features required by agile manufacturing systems, for integrating a new net-
work device that was not foreseen in advance at run time.
Apart from the vendor specific TTEB solutions, TSN standardised by the IEEE may play
an important role in the future industrial network systems [31]. It suppose to offer
high performance by utilisation of precise clock synchronisation protocol (802.1AS)
and support for scheduled traffic (802.1Qbv). However, no engineering or a scheduling
framework is specified in the current set of TSN amendments where such communica-
tion schedule could be calculated.
There are few scientific works dealing with the scheduling problem of TTEB systems,
which consider existing industrial requirements in terms of network topologies, com-
munication patterns or reliability. However, according to the author’s best knowledge,
no contributions exists that try to achieve the highest performance in terms of real-
time and reliability, while keeping the possibility to perform system reconfiguration at
runtime. In the dissertation of Dopatka [32], a framework for scheduling of communi-
cation based on graph colouring theory was developed. The assumption, however, was
that the system remains static, and the time necessary to calculate the communication
schedule is not of major importance. In his work, he did not considered redundancy to
improve the communication reliability. Hanzalek et al. in [30] also developed a sche-
duling framework that supports scheduling of PROFINET IRT protocol. The flexibility
and communication reliability were also not considered in their contribution. There are
also several contributions of Kopetz et al. about designing of Time Triggered communi-
cation systems that achieve high-time performance and reliability [33][9][34]. However
according to Poledna et al., the current deterministic systems based on TTEB are still
lacking flexibility features for performing online reconfiguration [35].
1.2 Objective
The main aim of this dissertation is to develop a scheduling framework for TDMA based
TTEB systems that considers typical industrial topologies (including redundancy) and
traffic characteristics, and is able to calculate a valid and possibly good schedule in a
short and deterministic time. Additionally, the objective is to develop a concept that
allows to automatically reconfigure TDMA based TTEB system at run-time, without the
necessity of stopping the running real-time communication and hindering its function-
ality.
Due to fast and deterministic scheduling approach the overall engineering time of
TTEB systems can be decreased. Furthermore, the scheduling algorithm can be also
executed on a resource limited hardware such as PLC. This will give the possibility that
in case of some changes, such as swapping of ports at a switch or exchanging network
device, the ”light“ version of the engineering (including only scheduling) can be per-
1.3. Outline of the document 5
formed directly at the PLC, without the necessity of performing manual efforts in an
engineering tool. This requirement is also induced by the TSN standard, that does not
have any additional engineering tool, where a communication schedule could be cal-
culated. Therefore, in this work it is assumed that the scheduling task will be overtaken
by one of the ordinary network members in a defined network segment.
The theoretical analysis of the proposed algorithms and methods will provide the ev-
idence about their efficiency as well as low complexity. The practical implementation
will help to asses theirs performance and allow to validate proposed methods. Further-
more, practical tests provide preliminary results of the proposed mechanisms and allow
to prove defined theories. Additional simulative experiments will complement the in-
vestigation with auxiliary results that were not possible or were difficult to obtain with
the existing hardware.
This chapter provides a general state of the art analysis that is relevant for this work.
Section 2.1 gives an overview about the production systems and provides deep insights
into NCSs build upon TTEB systems. Furthermore, in section 2.2 requirements in terms
of real-time and reliability that have to be fulfilled by INS are discussed and summarised.
Afterwards, in section 2.3, existing concepts to achieve deterministic and reliable real-
time communication are described. Section 2.4 elaborates on the engineering process
and all steps that have to be performed in order to bring an INS into work. Finally in
section 2.6 the problem statement of this dissertation is formulated and in section 2.7
the solution approach is briefly described.
in several EU projects, such as IoT-a [36], IoT@Work [37] and many others that can be
found in the European Research Cluster of the Internet of Things [38]. The other im-
portant aspect is to make “things” smart or “intelligent”, e.g. through self diagnosis,
self optimisation, etc. This can be achieved by extending the “physical world” (devices,
machines, tools, etc.) with algorithms and methods that run in a “cyber world” and in-
teract with the “physical world”, thus forming a new concept of Cyber Physical System
(CPS) [39][40]. All these concepts are part of the so called fourth industrial revolution
(Industry 4.0) [41][42][43].
Data size
Flexibility
ERP
MES/
SCADA
Timeliness
Reliability
Co
Networke ntrol System /
d Contro
l System
(NCS)
Technic
al Proc
ess (TP
)
liness, amount of data sent in one data packet or communication reliability. Level of
available flexibility is also different, which in this work is measured as the effort related
to integration of a new network component with the existing network infrastructure. In
case of vertical exchange, the most crucial aspect is the interoperability between differ-
ent systems located at different levels.
The focus in this dissertation is the lowest level of the automation pyramid that di-
rectly interacts with the TP. It has the most demanding requirements in terms of short
latency and jitter (defined as timeliness and described in section 2.2.1) and communi-
cation reliability (described in section 2.2.2). However the level of flexibility at this level
is rather limited.
• Sensors are units responsible for monitoring of variables of the TP in time, con-
verting the physical sizes, such as: speed, torque, temperature, pressure, etc. into
a electrical signal, thus creating a “digital picture” of the TP.
pends on the dynamics of the TP. For instance printing machines, having a high
process dynamics (motion control application), require short cycle times, i.e. be-
low 1ms with the jitter less than 1μs [50], while processes such as beer brewing
(process control application), require cycle times in the range of 10ms [50]. In
the industrial environment, the role of controller is typically performed by a spe-
cially designed Programmable Logic Controllers (PLCs) and therefore from now
on, this terms will be used to described such control units.
• Actuators are responsible to interact with the TP by changing digital signal into
energy or force using hydraulics, pneumatics, electric current, magnetics or me-
chanics in order to control it.
• IO devices (IOD) also called remote IO, are currently widely used to communi-
cate with multiple sensors and actuators, which are distributed in the field. In
comparison to older centralised systems, in which sensors and actuators are di-
rectly connected with a PLC, they allow to save a lot of wiring efforts, simplify the
system, thus save time and money. Communication between PLC and sensors
and actuators is performed using one physical interface (e.g. Ethernet) and one
communication protocol (e.g. EtherCAT).
These components, typically exchange small amount of data by using defined cycles
times [51] forming a Control System (CS). Two major classes of control systems can be
distinguished. These are open loop control and closed loop control systems, as shown
in figure 2.3.
Input Output
PLC actuator TP
Input Output +
PLC actuator TP -
sensor
a) b)
Figure 2.3: Control System a) Open Loop System, b) Closed Loop System with Feedback
the TP.
Today, components such as PLCs, IODs, sensors and actuators are highly distributed
among the production systems, thus there is a need to span them together with the
communication network. Such control systems are called Networked Control System
(NCS), where network becomes a part of the control loop, see figure 2.4. It is important
Input Output
PLC actuator TP
+
- Network IO Device
sensor
to mention that the network used in the control loop has to fulfil certain requirements
in terms of timelines and reliability. Several proprietary and standardised network tech-
nologies and protocols have been developed [52] that can be used to build NCSs. They
were explicitly designed to fulfil requirements of a large variety of industrial applica-
tions. A detailed description about the requirements is given in the following section.
Discussion about the way how protocols were designed in order to fulfil these require-
ments is given in section 2.3.
Actuator TP
Set point Output
HMI + PLC Heater Boiler
Network IO Device
Feedback
Temp. sensor
be performed cyclically. In order to be able to control the boiler in real-time, all tasks
performed by the NCS are time constrained. It means that tasks, such as data collection
from sensors, data processing by the PLC and data distribution to the actuators have
to be performed in a defined time window, see figure 2.6. This time window defines its
time window
time
release time (r i ) for each task, e.g. in case of data collection, defines the earliest time
when the sensor data is ready to be sent. Each task have also its deadline (d i ), which is
defined as the latest time, where a task has to be accomplished. In case of data collec-
tion, it would describe the latest time a sensor data has to be delivered to the PLC. Only
in such a way, it is possible to achieve appropriate reaction time of NCS over the TP. The
following subchapters will discuss the most important characteristic parameters that
has to be considered in designing of NCS. The focus is put on the INS that has to allow
to fulfil certain real-time as well as communication reliability requirements as a part of
the NCS. The next chapter discusses the timing requirements of the NCS in more detail.
tS1 message 1
message 1 Td1
tR1
tSn message N
tRn Tdn
message N
system. The transmission latency Td can be calculated by taking two time stamps, one
on the sender side at the time of sending the message (tS i ) and one at the reception
time of the message (tRi ). By the subtraction of tRi and tS i we obtain the Td . It is not
possible to exclude the network latency completely from the data delivery, however it is
important to keep the latency bounded. The parameter that describes the time differ-
ence between the worst case (Tdmax ) and the best case transmission latency is defined
Another random document with
no related content on Scribd:
concerned with Mr. Linder, the excellent agricultural inspector of the
Lindi municipality, to whom I owe many valuable suggestions, and
who, on account of his thorough acquaintance with this very district,
had originally been selected as my companion. Linder rendered
splendid service in suppressing the rebellion: while any action on the
part of the Field Force was still entirely out of the question, he had
already, with a small detachment of police, repulsed numerous
attacks of the rebels, and ultimately sustained a serious wound. But
while decorations have been simply raining down on the Navy and
the Schutztruppe, Bwana Linda still walks among mortals without a
single order. He is, however, a philosopher as well as a hero.
The song runs as follows:—
Ulendo wa Linda (er); pa kwenda ku Masasi na gumiri chikuo: mkasálile
mbwana mkubwa ngondo jaiche nand autwiche lunga yangadye. Mkasálile akida
Matora: ngondo jaiche na gombel(r)e lilōmbe. Tukujir(l)a Masasi; Mwera kupita
mchikasa mpaka pe Lindi. Ne wapere rukhsa. Yendeye ku mangwenu; mkapānde
mapemba.
The translation is as follows:—
“The journey of Linder, when he went to Masasi, and I shouted
with a shouting.—‘Tell the Bwana Mkubwa, war has come, and I ran
away without looking back. Tell the akida Matora, (that) the war has
come, and I have beaten the great-drum.’ Then we went to Masasi,
the Wamwera are beaten and go as far as Lindi, and they get
permission. ‘Go to your homes, and plant Mapemba (sorghum).’”
This is delivered in very quick recitative, and relates in a few words
the history of the whole campaign, of course making the singer the
central point. Mr. Linder comes to Masasi in the course of one of his
official tours, his principal duty being to ascertain whether the local
headmen have cultivated the various crops prescribed by
government. There the loyal Likoswe of course hastens to him and
warns him of impending hostilities on the part of the Wamwera.
Linder in his turn sends word to the District Commissioner at Lindi,
and at the same time despatches Likoswe with an urgent message to
Matola’s. Likoswe, on arriving, beats the war-drum (lilombe),
Matola’s warriors immediately hasten to the spot, six hundred men
with guns and many more with spears, bows and arrows, and the
chief marches on Masasi, to take the Wamwera in the rear. It is
related as a fact that Seliman Mamba and his subordinates had each,
at the beginning of the rising when their hopes were highest and they
already saw the Germans driven into the sea, fixed on a house at
Lindi with all its contents as his own share of the spoil. Possibly, the
line about the enemy’s going back to Lindi refers to these unrealised
plans. Matola, I believe, lost about forty men in fighting the rebels,
but certainly did not drive them back to Lindi. The last sentence
relates to the conclusion of peace:—the vanquished are pardoned,
and directed to go home quietly and plant their gardens once more.
At last the hunter changes his tactics. The dancer is, in fact, a
hunter, and not only that, but a very successful elephant-hunter; and
having just killed a large elephant, he is celebrating this deed of
prowess before the assembled inhabitants of his native village, just as
he does after his return from the actual hunt. Here, too, the people
have collected from far and near to see this celebrity, and to admire
his skill in the dance. His performance becomes more and more
vivacious—he no longer remains on one spot but trips forward, first
in a straight line, then in a zig-zag. At last he revolves in a circle,
moving round with short, cautious jumps, and all the time keeping
up the movements of his arms and hips without a moment’s
intermission. After one more rapid trip round the circle and a frantic
vibration of the whole body, the dancer stands still, breathing deeply.
This kind of dance is too peculiar, too divergent from all European
standards for us to judge of it critically according to the rules of art. I
had expected a pantomimic representation of an elephant-hunt, or at
least of the stalking and killing of the game, and I must confess that I
can find nothing in the performance which seems to have any such
reference, and must confine myself to admiring the incredible
dexterity shown by this acrobat in setting all his muscles a-quiver. I
have no sooner got a fresh film ready, than a second dancer has
appeared on the scene, whose action is still more curious and
perplexing. At first one sees nothing but a confused mass of green
leaves rolling and writhing on the ground in convulsive motions.
After a while, this resolves itself into a man much like the previous
one, except that his costume is much more voluminous. He quivers
in a masterly manner and shows as much staying power as his
predecessor; but his chief strength lies in his legs, whose suppleness
and power of assuming the most grotesque attitudes are nothing
short of marvellous. When he has exhausted his repertoire and made
way for a third performer, we at last get the expected pantomime.
Stooping as if for a spring, the hunter creeps up, noiselessly, making
use of every bit of cover, to stalk the elephant, whose scent is
exceedingly keen. At last the goal is reached—swiftly, but as
noiselessly as the hunter, the quarry, represented by another man,
has slipped into the arena, and squatted down, and the hunter circles
round him in diminishing spirals. We expect the deadly shot, but it
does not come off, and the third dancer, quite regardless of the
elephant he is supposed to represent, begins to “triumph” in
precisely the same way as the two others, practising highly artistic
short steps, swaying his hips and flourishing his arms. “Bassi”—
(finished,) I exclaim, as the last of my three films whizzes off the reel.
Quite in contrast to these are the typical unyago dances of the
Wayao. There seems to be a great variety of these; but so far I have
only seen two at Chingulungulu, a masewe, so called from the rattles
worn, as already mentioned, on the legs and feet, and a luwanja.
Both are essentially the same in character. The primitive xylophone
of the Makua hunting-dance is here replaced by a complete band of
drums, of the most various shapes and sizes. A certain musical
faculty inherent in the race is evidenced by the fact that the
musicians take care to tune up before the dance begins. Each beats
his own drum, listening carefully to hear whether it is in tune with
the rest, and if not, hurries away to the nearest hut and comes back
with a brand from the hearth and a large bundle of dry grass. The
grass is heaped on the ground and set on fire, and then every drum is
held with the open end over it, for a longer or shorter time—some for
a few seconds only, some for half a minute or more—the pitch being
tested by striking from time to time. At last all the skins are
sufficiently tense and the drumming begins.
A YAO DRESSED FOR
THE MASEWE DANCE