Download the full version of the textbook now at textbookfull.

com

Recent Advances in Computational

Optimization: Results of the Workshop on
“Computational Optimization” and “Numerical
Search and Optimization” 2018 Stefka Fidanova
https://textbookfull.com/product/recent-advances-
in-computational-optimization-results-of-the-
workshop-on-computational-optimization-and-
numerical-search-and-optimization-2018-stefka-
fidanova/

Explore and download more textbook at https://textbookfull.com

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

The ShortTube 80 Telescope A User s Guide Neil T. English

https://textbookfull.com/product/the-shorttube-80-telescope-a-user-s-
guide-neil-t-english/

textbookfull.com

Essentials of Hypertension: The 120/80 paradigm 1st

Edition Flávio Danni Fuchs (Auth.)

https://textbookfull.com/product/essentials-of-hypertension-
the-120-80-paradigm-1st-edition-flavio-danni-fuchs-auth/

textbookfull.com

Sew Your Own Wardrobe More Than 80 Techniques 1st Edition

Alison Smith

https://textbookfull.com/product/sew-your-own-wardrobe-more-
than-80-techniques-1st-edition-alison-smith/

textbookfull.com

Building Hybrid Android Apps with Java and JavaScript

Applying Native Device APIs Japplying Native Device Apis
1st Edition Gok Nizamettin Khanna Nitin
https://textbookfull.com/product/building-hybrid-android-apps-with-
java-and-javascript-applying-native-device-apis-japplying-native-
device-apis-1st-edition-gok-nizamettin-khanna-nitin/
textbookfull.com
Public Debt Dynamics of Europe and the US 1st Edition
Dimitris N. Chorafas (Auth.)

https://textbookfull.com/product/public-debt-dynamics-of-europe-and-
the-us-1st-edition-dimitris-n-chorafas-auth/

textbookfull.com

The Hundred Year Walk An Armenian Odyssey First Edition

Mackeen

https://textbookfull.com/product/the-hundred-year-walk-an-armenian-
odyssey-first-edition-mackeen/

textbookfull.com

Effective Project Management : Traditional, Agile, Extreme

Wysocki

https://textbookfull.com/product/effective-project-management-
traditional-agile-extreme-wysocki/

textbookfull.com

Gardens Illustrated Pot of Style 40 Beautiful Planting

Ideas for Containers Gardens Illustrated

https://textbookfull.com/product/gardens-illustrated-pot-of-
style-40-beautiful-planting-ideas-for-containers-gardens-illustrated/

textbookfull.com

Public Budgeting Reform in China Theory and Practice 1st

Edition Xiaonan Liu (Eds.)

https://textbookfull.com/product/public-budgeting-reform-in-china-
theory-and-practice-1st-edition-xiaonan-liu-eds/

textbookfull.com
Empirical Research in Statistics Education 1st Edition
Andreas Eichler

https://textbookfull.com/product/empirical-research-in-statistics-
education-1st-edition-andreas-eichler/

textbookfull.com
Studies in Computational Intelligence 838

Stefka Fidanova Editor

Recent Advances
in Computational
Optimization
Results of the Workshop on
“Computational Optimization” and
“Numerical Search and Optimization”
2018
Studies in Computational Intelligence

Volume 838

Series Editor
Janusz Kacprzyk, Polish Academy of Sciences, Warsaw, Poland
The series “Studies in Computational Intelligence” (SCI) publishes new develop-
ments and advances in the various areas of computational intelligence—quickly and
with a high quality. The intent is to cover the theory, applications, and design
methods of computational intelligence, as embedded in the fields of engineering,
computer science, physics and life sciences, as well as the methodologies behind
them. The series contains monographs, lecture notes and edited volumes in
computational intelligence spanning the areas of neural networks, connectionist
systems, genetic algorithms, evolutionary computation, artificial intelligence,
cellular automata, self-organizing systems, soft computing, fuzzy systems, and
hybrid intelligent systems. Of particular value to both the contributors and the
readership are the short publication timeframe and the world-wide distribution,
which enable both wide and rapid dissemination of research output.
The books of this series are submitted to indexing to Web of Science,
EI-Compendex, DBLP, SCOPUS, Google Scholar and Springerlink.

More information about this series at http://www.springer.com/series/7092

Stefka Fidanova
Editor

Recent Advances
in Computational
Optimization
Results of the Workshop on “Computational
Optimization” and “Numerical Search
and Optimization” 2018

123
Editor
Stefka Fidanova
Parallel Algorithms
Institute of Information and Communication
Technologies, Bulgarian Academy
of Sciences
Sofia, Bulgaria

ISSN 1860-949X ISSN 1860-9503 (electronic)

Studies in Computational Intelligence
ISBN 978-3-030-22722-7 ISBN 978-3-030-22723-4 (eBook)
https://doi.org/10.1007/978-3-030-22723-4
© Springer Nature Switzerland AG 2020
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, expressed 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.

This Springer imprint is published by the registered company Springer Nature Switzerland AG
The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland
Organization

Workshop on Computational Optimization (WCO 2018) is organized in the

framework of Federated Conference on Computer Science and Information Systems
FedCSIS—2018 and Numerical Search and Optimization part of the Numerical
Methods and Application Conference–2018.

Conference Co-chairs

For WCO
Stefka Fidanova, IICT-BAS (Bulgaria)
Antonio Mucherino, IRISA (Rennes, France)
Daniela Zaharie, West University of Timisoara (Romania)
For NSO
Stefka Fidanova, IICT-BAS (Bulgaria)
Gabriel Luque, University of Malaga (Spain)
Kalin Penev, Southampton Solent University (UK)

Program Committee

Bonates, Tibérius, Universidade Federal do Ceará, Brazil

Breaban, Mihaela, University of Iasi, Romania
Chira, Camelia, Technical University of Cluj-Napoca, Romania
Gonçalves, Douglas, Universidade Federal de Santa Catarina, Brazil
Hosobe, Hiroshi, National Institute of Informatics, Japan
Iiduka, Hideaki, Kyushu Institute of Technology, Japan
Lavor, Carlile, IMECC-UNICAMP, Campinas, Brazil
Marinov, Pencho, Bulgarian Academy of Science, Bulgaria
Micota, Flavia, West University Timisoara, Romania

v
vi Organization

Muscalagiu, Ionel, Politehnica University Timisoara, Romania

Parsopoulos, Konstantinos, University of Patras, Greece
Pintea, Camelia, Tehnical University Cluj-Napoca, Romania
Roeva, Olympia, Institute of Biophysics and Biomedical Engineering, Bulgaria
Siarry, Patrick, Universite Paris XII Val de Marne, France
Stefanov, Stefan, Neofit Rilski University, Bulgaria
Stoean, Ruxandra University of Craiova, Romania
Stoean, Catalin University of Craiova, Romania
Stuetzle, Tomas, Universite Libre de Bruxelles, Belgium
Tamir, Tami, The Interdisciplinary Center (IDC), Israel
Zilinskas, Antanas, Vilnius University, Lithuania
Preface

Many real-world problems arising in engineering, economics, medicine, and other

domains can be formulated as optimization tasks. Every day we solve optimization
problems. Optimization occurs in the minimizing time and cost or the maximization
of the profit, quality, and efficiency. Such problems are frequently characterized by
non-convex, non-differentiable, discontinuous, noisy or dynamic objective func-
tions, and constraints which ask for adequate computational methods.
This volume is a result of very vivid and fruitful discussions held during the
Workshop on Computational Optimization and Workshop on Numerical Search
and Optimization. The participants have agreed that the relevance of the conference
topic and quality of the contributions have clearly suggested that a more compre-
hensive collection of extended contributions devoted to the area would be very
welcome and would certainly contribute to a wider exposure and proliferation of the
field and ideas.
The volume includes important real problems like modeling of physical pro-
cesses, wildfire and natural hazards modeling, modeling metal nanostructures,
workforce planning, wireless network topology, parameter settings for controlling
different processes, extracting elements from video clips, and management of cloud
computing environment. Some of them can be solved by applying traditional
numerical methods, but others need huge amount of computational resources.
Therefore for them are more appropriate to develop an algorithm based on some
metaheuristic methods like evolutionary computation, ant colony optimization,
particle swarm optimization, constraint programming, etc.

Sofia, Bulgaria Stefka Fidanova

March 2019 Co-Chair
WCO’2018

vii
Contents

Developing a Method for Measuring the Failover Times of First Hop

Redundancy Within Video Networks . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
Paul Bourne, Neville Palmer and Jan Skrabala
Desktop Application Developed by Open Source Tools
for Optimizations in Cases of Natural Hazards
and Field Response . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
Nina Dobrinkova and Stefan Stefanov
Data Optimizations on Kresna Fire (2017) as Inputs for WFA
Simulations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
Nina Dobrinkova
Solving Sorting of Rolling Stock Problems Utilizing Pseudochain
Structures in Graphs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
Jens Dörpinghaus and Rainer Schrader
InterCriteria Analysis of Different Hybrid Ant Colony Optimization
Algorithms for Workforce Planning . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
Stefka Fidanova, Olympia Roeva, Gabriel Luque and Marcin Paprzycki
Different InterCriteria Analysis of Variants of ACO algorithm
for Wireless Sensor Network Positioning . . . . . . . . . . . . . . . . . . . . . . . . 83
Olympia Roeva and Stefka Fidanova
Geometric Versus Arithmetic Difference and H1-Seminorm Versus
L2-Norm in Function Approximation . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
Stefan M. Filipov, Ivan D. Gospodinov, Atanas V. Atanassov
and Jordanka A. Angelova
Ellipsoidal Estimates of Reachable Sets for Nonlinear Control Systems
with Bilinear Uncertainty . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
Tatiana F. Filippova

ix
x Contents

Structural Instability of Gold and Bimetallic Nanowires Using Monte

Carlo Simulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133
Vladimir Myasnichenko, Nickolay Sdobnyakov, Leoneed Kirilov,
Rossen Mikhov and Stefka Fidanova
Manipulating Two-Dimensional Animations by Dynamical Distance
Geometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147
Antonio Mucherino
Index Matrices as a Cost Optimization Tool of Resource Provisioning
in Uncertain Cloud Computing Environment . . . . . . . . . . . . . . . . . . . . . 155
Velichka Traneva, Stoyan Tranev and Vassia Atanassova
Three-Dimensional Interval-Valued Intuitionistic Fuzzy Appointment
Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 181
Velichka Traneva, Vassia Atanassova and Stoyan Tranev
Logical Connectives Used in the First Bulgarian School Books
in Mathematics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 201
Velislava Stoykova
An Improved “Walk on Equations” Monte Carlo Algorithm
for Linear Algebraic Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 215
Venelin Todorov, Nikolay Ikonomov, Stoyan Apostolov, Ivan Dimov,
Rayna Georgieva and Yuri Dimitrov
Author Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 237
Visit https://textbookfull.com
now to explore a rich
collection of eBooks, textbook
and enjoy exciting offers!
Developing a Method for Measuring
the Failover Times of First Hop
Redundancy Within Video Networks

Paul Bourne , Neville Palmer and Jan Skrabala

Abstract IP protocols have been used to distribute compressed media over private
and public networks for a number of years. Recently the broadcast sector has started
to adopt IP technologies to transport real time media within and between their facil-
ities during production. However, the high bitrate of uncompressed media and its
sensitivity to latency and timing variations requires careful design of the network in
order to maintain quality of service. Connectionless protocols are commonly used,
which means that packet loss is of particular concern and redundant paths must be
provisioned with mechanisms to switch between them. This project develops and
critically analyses a method for measuring the effectiveness of first hop redundancy
protocols for broadcast video production networks. The aim extends previous work
[1] to recommend particular configurations to optimise networks and to provide a
method that broadcast engineers can use to verify performance. Cisco’s HSRP is
recommended with static routes configured for redundant paths. It is recommended
that the network is tested using a synthetic RTP stream with a low complexity packet
sniffer and NICs with hardware timestamps. Further work is identified including
ways to improve the accuracy of the results and to consider the impact of more
complex networks.

Keywords FHRP · HSRP · GLBP · VRRP · Network redundancy · Gateway

protocol · Failover · Network performance measurement · Video network

P. Bourne (B) · N. Palmer

Solent University, East Park Terrace, Southampton, Hampshire SO14 0YN, UK
e-mail: [email protected]
N. Palmer
e-mail: [email protected]
J. Skrabala
Vostron Ltd, 32B Castle Way, Southampton, Hampshire SO14 2AW, UK
e-mail: [email protected]

© Springer Nature Switzerland AG 2020 1

S. Fidanova (ed.), Recent Advances in Computational Optimization,
Studies in Computational Intelligence 838,
https://doi.org/10.1007/978-3-030-22723-4_1
2 P. Bourne et al.

1 Introduction

The broadcast industry is currently undergoing a step change as it replaces bespoke

network infrastructures based on the Serial Digital Interface (SDI) with commodity
enterprise equipment using Internet Protocol (IP). Consumers in the UK are supple-
menting traditional broadcast channels with online services; TVB Europe’s recent
outlook report [2] states that viewers prefer streaming services over offline down-
loads and continues that such services are predicted to grow by 9.1% each year over
the next five years. Meyer and Francis [3] suggest that the release of a number of IP
related standards throughout 2017, as well as native IP products, has also increased
broadcasters’ confidence in IP technologies to the point that they are now investing
heavily in IP infrastructures for playout and distribution. Broadcast services provider,
Timeline Television, recently unveiled the first all IP outside broadcast truck [4]. The
result is a desire for IP based infrastructures throughout the broadcast chain from
capture through to delivery.
There are many advantages in terms of cost and flexibility but maintaining Quality
of Service (QoS) for signals that are both time-sensitive and of high bandwidth can
be challenging. Broadcasters are used to fixed bandwidth links and bespoke health
monitoring mechanisms. In contrast network infrastructures have evolved with dif-
ferent requirements and have therefore developed significantly different techniques
and monitoring tools. In recent years there have been significant efforts to help the
sectors to converge, led by organisations such as the Broadcast Bridge and Alliance
for IP Media Solutions [5, 6].
Higher level applications are often used to monitor network services and create
reliable circuits using technologies such as Software Defined Networks (SDN) and
Mulit Protocol Label Switching (MPLS). These tools are usually provided by third-
party vendors based on proprietary intellectual property and produce a high degree
of abstraction [7–9]. There are standard redundancy protocols available that can be
configured to respond to changes in network performance. However, monitoring
QoS in IP networks has proved to be difficult for broadcasters as the metrics built
into enterprise equipment tend to be geared towards the operational requirements
of generic data networks. The levels of abstraction and limited detail available may
leave broadcast engineers wary of the technologies and either less likely to adopt
them or to wildly overprovision the networks. The broadcast industry recognises the
need for network engineers who understand the requirements of media signal flows
and are starting to understand that a detailed understanding of networks is required to
configure effective production LANs [10–12]. A review of current literature provides
limited guidance on the relative performance of redundancy protocols and there are
no procedures for measuring the performance of high availability networks.
This project aims to evaluate the effectiveness of existing network redundancy
protocols and their suitability for broadcast video networks with a focus on produc-
tion networks. The project will explore and evaluate methods of monitoring network
performance and recommend a test procedure to allow engineers to verify and opti-
mise their configurations.
Developing a Method for Measuring the Failover Times … 3

2 Background

IP networks are becoming increasing important throughout the broadcast video work-
flow from production through to distribution. Value can be added to the signal chain
via the flexibility of dynamically routed signals and format agnostic transport proto-
cols. There has been significant uptake of services distributed using IP technologies
such as Over the Top (OTT) and Internet Protocol Television (IPTV) [13, 14]. OTT
services such as Netflix and BBC iPlayer are delivered over the public Internet with-
out the need for proprietary devices whereas IPTV services such as Virgin Media
use managed networks. Due to the nature of the networks OTT is usually deliv-
ered using Hyper Text Transport Protocol (HTTP) using the connection-orientated
Transmission Control Protocol (TCP). With the use of a buffer, this provides bene-
fits such as retransmission but at the expense of latency. IPTV or real-time signals
within a production environment are more likely to use Realtime Transport Protocol
(RTP) over the connectionless User Datagram Protocol (UDP). RTP is used by many
media-centric protocols as a transport mechanism to provide sequence numbers and
timestamps with minimal overhead. The information provided within the headers
may be used by high level applications to improve QoS and further feedback is usu-
ally exchanged between participants using the RTP Control Protocol (RTCP) [15].
The small receive buffer and lack of a retransmission mechanism means that RTP
is vulnerable to packet-loss and necessitates a rapid failover mechanism to maintain
resilience using a secondary path.
First Hop Redundancy Protocols (FHRP) provide an essential tool for increas-
ing availability in critical switched IP networks. They provide a mechanism for fast
failover to the next hop from a primary path to a secondary path within a group of
backup routers. The process is faster than waiting for spanning tree or dynamic rout-
ing protocols to converge on a new path due to the limited scope and pre-configuration
of the FHRP. Essentially two or more routers are able to share the default gateway
at OSI layer 3, which provides an alternative route or may even be used for rudi-
mentary load balancing. This does necessitate the use of multilayer switches but can
be implemented at the Access Layer or Distribution Layer [16, 17]. YanHua and
WeiZhe [18] have shown that such protocols are suitable for use within cable tele-
vision IP networks when combined with device redundancy, although they caution
against diminishing returns as the network complexity increases.
Cisco has developed two major proprietary protocols Hot Standby Routing Proto-
col (HSRP) and Gateway Load Balancing Protocol (GLBP). Another common proto-
col is the Virtual Router Redundancy Protocol (VRRP), which is available as an open
IEFT standard RFC5798. It is similar to HSRP in operation but not compatible [19,
20]. Other equipment vendors also have proprietary protocols; Juniper Networks
has NetScreen Redundancy Protocol (NSRP), Avaya has Routed Split Multi-link
Trunking (R-SMLT) and Extreme Networks has Extreme Standby Routing Protocol
(ESRP) [21–24]. Broadcast installations usually combine best-of-breed equipment
and are generally multivendor environments built on open standards. However Cisco
have a dominant position in the switching and routing market with over 50% of the
4 P. Bourne et al.

worldwide market share in 2016 [25]. As such this investigation will focus on HSRP,
GLBP and VRRP. Common Address Redundancy Protocol (CARP) was considered
but relies on the Berkeley Software Distribution so is not practical on many platforms
[26].
HSRP is configured for an interface using the standby command and allows the
user to configure a virtual gateway for the connected hosts to use. Priorities are
specified such that an active router is allocated with one or more standby routers
sharing the virtual address with the active router as shown by Fig. 1. Packets are
forwarded based on an IP/MAC address pair and standby routers monitor the status
of the active router to promote a backup router in the case of a link failure on the active
path. Tracking objects can be used to monitor interfaces or Service Level Agreement
(SLA) tracking can monitor connectivity beyond the first hop. Either method can
update the router priorities to determine the active path. Different priorities can also
be assigned to different Virtual Local Area Networks (VLAN) to implement basic
load balancing, although this may become unwieldy on large networks [17, 27].
GLBP uses multiple gateways simultaneously, which enables more effective load
balancing and therefore uses all the bandwidth within the topology. Routers within a
GLBP group may be the Active Virtual Gateway (AVG), an Active Virtual Forwarder
(AVF) or the Standby Virtual Gateway (SVG). The AVG assigns virtual MAC address
to the other group members. Up to four AVFs, including the AVG, are able to forward
packets and the SVG is ready to take over from the AVG based on a similar priority
system to HSRP with decrements based on tracking objects. GLBP is implemented
on an interface using the glbp command and load balancing can be achieved within
the group by assigning packets to the MAC addresses of the AVFs via an equal
round-robin, by weighting certain paths or based on the host [17, 28].

Fig. 1 A typical HSRP configuration with the host using a common virtual gateway IP address
Developing a Method for Measuring the Failover Times … 5

VRRP is very similar to HSRP in that is uses a single virtual gateway that is shared
between a master and one or more backup router. It is implemented on the interface
using the vrrp command and supports object tracking to determine failures. Fewer
IP addresses may be used by VRRP than HSRP as the physical IP address for the
master router may also be used as the virtual IP address [29].
Several articles have been published that outline the configuration options for
HSRP, GLBP and VRRP [20, 30–32]. These discuss how the options affect the
underlying algorithm but there appears to be little guidance on how to optimise the
parameters or how the common protocols compare under similar conditions. Ibrahimi
et al. [32] and Rahman et al. [33] demonstrate a basic method, using continuous
Internet Control Message Protocol (ICMP) echo requests to show the duration of a
link failure. Pavlik et al. [17] show a more accurate method to determine the interval
between missing and restored replies using timestamps from a packet sniffer. None of
the studies investigate the time taken to restore a link after the primary link recovers.
QoS describes the technical performance of a network, which can usually be
objectively quantified and measured at specific points within the network topology.
Common parameters include latency, jitter and packet loss but bandwidth and packet
reordering are also important within video networks due to the temporal sensitivity
and inter-packet dependencies of media flows [34, 35]. Latency is the time it takes
for packets to reach their destination whereas jitter is the variance in the inter-packet
latency. Latency is often acceptable when there is minimal jitter or packet loss,
although it can be a problematic in live broadcasting when disparate sources are
combined such as when conducting interviews. Packet loss tends to be caused by
excessive latency or jitter that causes buffers to underflow and tends to result in
audio or video frames freezing or degrading. Receive buffers can reduce jitter and
therefore packet loss but at the expense of latency [36]. Bandwidth is of concern
within production facilities due to the high data rates of video at 2.97 Gbps for a
single uncompressed High Definition (HD) stream. Gharai, Perkins and Lehman [37]
have noted that in video networks, packet reordering should be treated on an equal
footing with packet loss and the European Broadcast Union (EBU) recommends that a
receiver should not have to accommodate packets out of order by more than 10 places.
Packet reordering is often caused during redundancy switches. Cisco recommends
that for video networks latency should be less than 300 ms, jitter less than 50 ms and
packet loss of less than 0.5% [38] although production environments may have to
work to stricter limits for control and monitoring with latency as low as 10 ms and
virtually no packet loss [39, 40].

3 Test Method

How to measure the effectiveness of FHRPs is one of the major objectives of the
research and a clear and accessible method should be one of the outputs. Previous
studies of the quality of video over IP have made a distinction between the QoS
provision from the network and the QoS provision from the media application [36,
6 P. Bourne et al.

41]. Studies at the application layer have tended to focus on comparing image qual-
ity by detecting artefacts such as blocking and blurring or calculating Peak Signal
to Noise Ratio (PSNR) using bespoke measurement applications. These are very
dependent on the configuration of the codec and often require reliable side-channels
to make mathematical comparisons with a reference stream. This study will instead
focus on the QoS of the network, which Tao, Apostolopoulos and Guerin [41] state
is largely driven by packet loss, delay and jitter. This may be achieved by generating
and injecting a stream of timestamped packets into a test network to be captured and
analysed at a receiver. Previous studies into FHRPs demonstrate a method to test
availability using a continuous stream of Internet Control Message Protocol (ICMP)
echo requests to determine the duration of a link failure [32, 33]. Alternatively a
packet sniffer can be used to observe the interval between missing echo relies as
demonstrated by Pavlik et al. [17]. Essentially packet drop and the unavailability
interval provide the same information as the packet drop is a function of the interval
and bitrate.
There are several network simulation packages that may be used to rapidly anal-
yse the behaviour of different configurations. Common simulators include Cisco’s
Packet Tracer, open source application GNS3 and Riverbed Modeler. For these to
produce accurate results, they require detailed implementations of the software and
protocols running on the network devices as well as accurate models of the hardware
performance. The level of difficulty required to accurately reproduce the test envi-
ronment is not necessary where real equipment is available to provide accurate real
time results.
A testbed was created to emulate a typical spine-and-leaf network architecture as
recommended for high available networks. The topology is shown by Fig. 2.
The source had an edge ‘customer’ router, which connected to their Local Area
Net-work (LAN). This had redundant links to an external network via primary and
secondary routers, which were connected to the ‘main’ router at the destination on a
WAN. The switch simplified the configuration of the customer router by removing
the need for two interfaces on the same subnet. The primary link utilised a Gigabit
Ethernet connection whereas the secondary was only Fast Ethernet. Secondary links
are often metered in practice so load balancing was not implemented. To simulate a
link failure, the interface G0/0 was shut down on the primary router with a tracking
object to promote the secondary link based on the line-protocol state. The FHRPs
were configured to decrement the priority of the primary router below that of the
secondary router if the line protocol went down; this causes the secondary router to
pre-empt the primary router and traffic would be rerouted. An alternative method
would be to check the reachability of the loopback interface on the destination net-
work using ICMP requests to decrement the priority upon failure. This would be a
more meaningful detection method within a real network but the non-deterministic
nature of packet generation and propagation may distort the results, which should be
focused on the responsiveness of the redundancy protocol itself.
All routes were statically defined within the routers to prevent the dynamic rout-
ing protocols from interfering with the results. The FHRPs were configured on the
primary and secondary routers with a default static route on the customer router
Developing a Method for Measuring the Failover Times … 7

Fig. 2 Topology for FHRP experiment

pointing to the virtual IP address. A default and more rapid configuration was tested
for each FHRP using the parameters shown by Table 1. By default HSRP sets the
time between hello messages to 3 s and the hold time to be 10 s. The hold time is the
interval after which the active router is declared to be down once hello messages are
not acknowledged. The hold timer must be greater than the hello timer—usually at
least three—to avoid the active router being declared down between hello messages
[42, 43]. GLBP uses the same system of hello and hold timers with identical default
and minimum values [44]. HSRP and GLBP timers can be set with millisecond gran-
ularity but a shorter timer increases the amount of overhead traffic and can cause the
system to behave erratically as it becomes too sensitive [34]. For the rapid test the
hello timer will be set to 1 s and the hold timer to 3 s; this will allow the effect
of smaller timers to be observed without creating instability. VRRP advertisements
are sent every second by default with a failover delay calculated by three times the
advertisement interval plus the router’s skew time. The skew time is based on the
8 P. Bourne et al.

Table 1 Active variables for FHRP experiment

Hello interval (s) Hold timer (s) Pre-emption delay (s) Priority
(backup/master)
HSRP D 3 10 0 –
HSRP R 1 3 0 –
GLBP D 3 10 30 –
GLBP R 1 3 3 –
VRRP D 1 – 0 100/110
VRRP R 1 – 0 253/254

inverse of the priority. The standard doesn’t include shorter advertisement intervals
but Cisco has implemented this in their IOS down to 20 ms [30, 45]. The priorities
were configured as high as possible for the rapid test to reduce the skew time. It
should be noted that pre-emption delay should normally be configured to allow for
the boot time of the equipment [34]. Each configuration was tested ten times.
Two methods were employed for measuring the interval between the interface
changing state and the FHRP responding. A third method that would track missing
ICMP echo replies was considered and trialed during earlier studies [1]; the results
from this would be more comparable to existing studies but the flow was found to
be too simplistic and imprecise to represent media streams.
Method 1: The packet loss was measured using a test stream and packet sniffer.
This relies on the accuracy of the packet generator as well as the timestamps within
the captured stream. VLC Media Player was used to generate a stream of colour bars
and tone from an MP4 file. netsniff-ng was used to capture traffic for analysis using
Wireshark’s RTP dissector. The interval between missing sequence numbers could
be used to determine the packet loss during failover or restoration. The jitter on the
captured packets was also measured to determine the inter-packet arrival variance
and therefore the precision to which the time interval could be reliably calculated.
Method 2: Timestamped debug messages on the routers were used to report the
interface failure and pre-emption of a new router. NTP was used to synchronise
the internal clocks so that the timestamps from different devices could be compared.
The accuracy of the log timestamps was investigated by sending 1000 PING requests
from the source PC to the primary router with 100 ms interval. The SPAN feature on
the primary router was used to copy packets from the G0/1 interface to a netsniff-ng
packet capture on the SPAN PC. Logging was configured on the primary router to
millisecond accuracy using an Access Control List (ACL) to also log the arrival of
the ICMP echo requests. The timestamps on the packet capture were compared to
the logging timestamps to indicate any discrepancy with the logging process. The
jitter inherent on the network was also measured using the Iperf utility.
Developing a Method for Measuring the Failover Times … 9

4 Results

Of the 1000 PING requests sent, the maximum deviation from the intended 100 ms
interval was ± 4 ms within the raw Cisco timestamps. The PING utility and source
interface will introduce jitter so the figures were adjusted based on the inter-packet
arrival intervals from the SPAN capture. This reduced the error slightly to +4 ms
and −3 ms. In total 31 packets were logged as having a different arrival time to the
SPAN data with an average absolute deviation of 2 ms. The round trip time reported
by the PING utility ranged from 1 to 2 ms with an average of 1.88 ms.
After running 10 cycles of the jitter measurements, the maximum jitter reported
was 570 µs, the minimum 26 µs and the mean average result was 82 µs with a
standard deviation of just 0.1.
The results from the experiment suggest that Cisco’s logging timestamps have
the potential to be accurate to within +4/−3 ms. The jitter added from the test
equipment was much lower than this, which suggests that the figures are reliable. This
is enough precision to compare FHRP but Accedian [46] suggest that inaccuracies in
timestamps approaching even a tenth of a millisecond are unsuitable for measuring
high performance SLA metrics. One concern is that the load on the router was
extremely low during this experiment as there was no significant traffic being routed
or other events for the router to log. As such this indicates the best accuracy of the
logging timestamp.
The minimum, maximum and mean average time have been presented for the
default and rapid configurations of each redundancy protocol. The failover times
are summarised in Table 2 and the restoration times in Table 3. The average jitter
calculated from the stream captures is included to provide an indication of the system
stability—jitter data has not been presented for the rapid configuration of VRRP as
it was consistently reported to be zero. The mean absolute differences between the
results from the two methods are shown as absolute value and percentages.
The mean failover and restoration delays as well as the mean number of packets
dropped are shown as radar plots for the default and rapid configuration in Fig. 3 and
Fig. 4 respectively. The metrics have been scaled to produce a plot whereby the outer
edge indicates the worst performance and the center indicates ideal performance.
It can be seen from Tables 2 and 3 that the two test methods produce similar results
for VRRP and GLBP but there is significant disparity for HSRP—the captures suggest
that the protocol is 83–88% faster than the logs report. Inspection of the packet
captures verifies that this figure is correct and suggests that the logging process
introduces a delay with this protocol. The results correlate well with the studies
completed by Pavlik et al. [17] although it is interesting to note that all of the protocols
were faster than in their study. This is likely to be due to the different topologies used
and will be particularly affected by the use of static routes in this study. Previous
studies have not analysed the restoration delays, which were also found to be fastest
with HSRP closely followed by VRRP.
The packet loss showed some particularly interesting results. The rapid failover
of HSRP led to the fewest losses whilst GLBP resulted in the most. Surprisingly
10 P. Bourne et al.

Table 2 FHRP failover PCap RTP sequence errors

delays in seconds from packet
capture and cisco logs Min Max Mean Jitter Dropped
HSRP D 0.325 2.576 1.527 0.069 29
HSRP R 0.313 1.026 0.655 0.069 9
VRRP D 3.090 3.937 3.461 0.071 76
VRRP R 2.244 2.874 2.602 – 79
GLBP D 30.705 38.928 33.425 0.069 670
GLBP R 3.913 4.894 4.332 0.071 678
Cisco logs Mean absolute
difference (%)
Min Max Mean
HSRP D 11.188 13.344 12.280 10.753 87.56
HSRP R 0.644 3.856 3.274 2.713 82.87
VRRP D 2.784 3.552 3.206 0.307 9.57
VRRP R 2.192 2.900 2.548 0.581 22.81
GLBP D 30.464 34.984 32.337 1.107 3.42
GLBP R 3.652 4.580 4.047 0.285 7.04

Table 3 FHRP restoration PCap RTP sequence errors

delays in seconds from packet
capture and cisco logs Min Max Mean Jitter Dropped
HSRP D 0.293 2.692 1.484 0.069 0
HSRP R 0.417 1.009 0.725 0.069 0
VRRP D 2.360 3.535 3.132 0.071 574
VRRP R 2.205 2.878 2.508 – 0
GLBP D 31.200 46.519 35.533 0.069 0
GLBP R 2.993 3.358 3.196 0.071 0
Cisco logs Mean absolute
difference (%)
Min Max Mean
HSRP D 0.368 2.660 1.436 0.083 5.77
HSRP R 0.156 0.944 0.633 0.081 12.83
VRRP D 2.840 3.532 3.166 0.168 5.31
VRRP R 2.288 2.924 2.698 0.239 8.88
GLBP D 31.108 46.512 34.654 0.874 2.52
GLBP R 3.100 3.196 3.128 0.103 3.28
Visit https://textbookfull.com
now to explore a rich
collection of eBooks, textbook
and enjoy exciting offers!
Developing a Method for Measuring the Failover Times … 11

Fig. 3 A comparison of protocol performance with default configurations

Fig. 4 A comparison of protocol performance with rapid configurations

12 P. Bourne et al.

GLBP resulted in slightly more loss with its rapid configuration even though the link
initially came back up more quickly. VRRP consistently dropped a significant number
of packets during restoration when configured with the default parameters—no other
configuration exhibited this behaviour including the rapid configuration of VRRP.
This is highly undesirable as it will result in an additional video glitch in the event
of a failover.

5 Conclusions

The initial tests logging ICMP echo requests suggested that timestamps could be
accurate to within ± 4 ms and that the jitter from low-level packet generation tools
would have minimal influence on the results. However, the disparity between the logs
and packet captures show that the timestamps from logs cannot be relied upon for
more complex protocols—this could be due to the interface between the protocol and
the logging procedure or due to the priority that is assigned to logging the protocol. It
is recommended that it is more accurate to capture an RTP stream to test the failover
delay.
Originally packEth was going to be used to generate an RTP stream. The thread-
ing used to generate the traffic is reported to be stable to microsecond resolution
[47] and Srivastava et al. [48, 49] found the tool to be more suitable than alter-
natives for testing high capacity links. However this tool doesn’t produce RTCP
messages, which caused the routers to drop the RTP stream. Instead VLC Media
Player was used, which had to encode real video frames rather than generating a
synthetic stream. The captures showed a throughput of approximately 20–30 pack-
ets per second, which didn’t give much resolution and the delta between consecutive
packets varied from approximately 0.04 to 250 ms, which is quite bursty. The resolu-
tion could be improved by increasing the data-rate at the source but the burstiness is
likely to be harder to solve. The jitter in the network was fairly consistent throughout
the experiment at around 70 ms. A video network would be expected to have jitter
below 30 ms to maintain QoS [50]. The high figure is likely to be due to using a
software application to generate the RTP stream. The worst case jitter originating
from the source and capture NICs themselves was measured to be just 0.57 ms. A
tool is needed to synthetically generate RTP streams with RTCP messaging so that
a high throughput can be generated with low jitter. This could be an extension to
packEth or a separate utility.
The results themselves suggest that HSRP is the most effective protocol and that
the rapid configuration improves its performance. This protocol is Cisco proprietary
so VRRP would have to be used within a multivendor environment with the config-
uration customised to avoid the packet loss observed using the default parameters.
This work could be extended by reducing the configured timers further to find the
limits of the protocol. As broadcasters are likely to use larger systems with best of
breed equipment, the effects of multiple standby devices and mixed-vendor equip-
ment should also be investigated.
Developing a Method for Measuring the Failover Times … 13

References

1. Bourne, P., et al.: Quantifying the effectiveness of first-hop redundancy protocols in IP networks.
In: Numerical Methods and Applications, NMA 2018. Lecture Notes in Computer Science,
vol 11189. Springer, Cham (2019)
2. Ramsey, C.: 2018 Mediatech Outlook. Newbay, s.l. (2018)
3. IP in 2018—A Growth in Market Confidence. https://www.tvtechnology.com/news/ip-in-2018-
a-growth-in-market-confidence. Accessed 27 Jan 2019
4. Timeline Television to Showcase 4 K-HDR IP OB Truck on SAM’s IBC 2017 Booth. https://
www.thebroadcastbridge.com/content/entry/9387/timeline-television-to-showcase-its-ip-4k-
hdr-ob-truck-on-sams-ibc-2017-boo. Accessed 21 Jan 2019
5. About—The Broadcast Bridge. https://www.thebroadcastbridge.com/about. Accessed 21 Jan
2019
6. Overview—AIMS Alliance. https://www.aimsalliance.org/overview/. Accessed 21 Jan 2019
7. Man, H., Li, Y.: Multi-stream video transport over MPLS networks. In: IEEE Workshop on
Multimedia Signal Processing, pp. 384–387 (2002)
8. Aria Networks Shows Off Self-Healing Video Distribution Using SDN + AI Combo.
http://www.aria-networks.com/news/aria-networks-self-healing-video-distribution-sdn-ai/.
Accessed 21 Jan 2019
9. Demystifying SDN: It’s an integral part of next-gen IP video transport. https://www.
tvtechnology.com/expertise/demystifying-sdn. Accessed 21 Jan 2019
10. TV, Vendors Struggle To Keep Up With Tech. https://tvnewscheck.com/article/114036/tv-
vendors-struggle-to-keep-up-with-tech/. Accessed 21 Jan 2019
11. What Broadcasters Can Learn From Telcos About IP. https://www.tvtechnology.com/
broadcast-engineering/what-broadcasters-can-learn-from-telcos-about-ip. Accessed 21 Jan
2019
12. Whether It’s Called IP or IT, Media Networks Are Different. https://www.thebroadcastbridge.
com/content/entry/7062/whether-its-called-ip-or-it-media-networks-are-different. Accessed
21 Jan 2019
13. Quarter of European OTT platforms target cross-border audiences. https://www.
broadbandtvnews.com/2017/11/15/quarter-of-european-ott-platforms-target-cross-border-
audiences/. Accessed 21 Jan 2019
14. The revolutionary transition to an all-IP video production workflow. https://www.ibc.
org/production/revolutionary-transition-to-all-ip-video-production-workflow/925.article.
Accessed 21 Jan 2019
15. Cheng, Y., et al.: From QoS to QoE: a tutorial on video quality assessment. IEEE Commun.
Surv. Tutor 17(20), 1126–1165 (2015)
16. CCNA Routing and Switching Practice and Study Guide: LAN Redundancy. http://www.
ciscopress.com/articles/article.asp?p=2204384&seqNum=4. Accessed 21 Jan 2019
17. Pavlik, J., et al.: Gateway redundancy protocols. In: 15th IEEE International Symposium on
Computational Intelligence and Informatics, pp. 459–464. IEEE, Budapest (2014)
18. YanHua, Z., WeiZhe, M.: The design of cable television IP access network based on hot standby
router protocol. In: International Conference on Image Analysis and Signal Processing, pp. 1–4.
IASP, Hangzhou (2012)
19. Virtual Router Redundancy Protocol (VRRP) Version 3 for IPv4 and IPv6. https://tools.ietf.
org/html/rfc5798. Accessed 21 Jan 2019
20. Oppenheimer, P.: Top-Down Network Design, 3rd edn. Cisco Press, Indianapolis (2011)
21. OpenBSD PF—Firewall Redundancy (CARP and pfsync). https://www.openbsd.org/faq/pf/
carp.html. Accessed 21 Jan 2019
22. NSRP overview. https://kb.juniper.net/InfoCenter/index?page=content&id=KB4263&cat=
HA_REDUNDANCY&actp=LIST. Accessed 21 Jan 2019
23. Routed split multilink trunking. https://worldwide.espacenet.com/publicationDetails/biblio?
FT=D&date=20050113&DB=&locale=en_EP&CC=US&NR=2005007951A1&KC=A1&
ND=1. Accessed 21 Jan 2019
14 P. Bourne et al.

24. ExtremeWare XOS Concepts Guide Software Version 11.3, p. 445. https://www.manualslib.
com/manual/511330/Extreme-Networks-Extremeware-Xos-Guide.html?page=1#manual.
Accessed 4 July 2018
25. Switching market stays steady as Cisco’s domination continues. https://www.cbronline.
com/data-centre/switching-market-stays-steady-as-ciscos-domination-continues-4993282/.
Accessed 21 Jan 2019
26. Chapter 28. Introduction to the Common Address Redundancy Protocol (CARP). https://www.
netbsd.org/docs/guide/en/chap-carp.html. Accessed 21 Jan 2019
27. Hot Standby Router Protocol Features and Functionality. https://www.cisco.com/c/en/us/
support/docs/ip/hot-standby-router-protocol-hsrp/9281-3.html. Accessed 21 Jan 2019
28. Cisco: First Hop Redundancy Protocols Configuration Guide. Cisco IOS Release 12.2SX.
Cisco, San Jose (2011)
29. VRRP Overview. https://www.juniper.net/documentation/en_US/junose15.1/topics/concept/
vrrp-overview.html. Accessed 21 Jan 2019
30. First Hop Redundancy Protocols Configuration Guide, Cisco IOS Release 15SY. https://www.
cisco.com/c/en/us/td/docs/ios-xml/ios/ipapp_fhrp/configuration/15-sy/fhp-15-sy-book/fhp-
vrrp.html. Accessed 21 Jan 2019
31. Configuring VRRP. https://www.juniper.net/documentation/en_US/junos/topics/example/
vrrp-configuring-example.html. Accessed 21 Jan 2019
32. Ibrahimi, M., et al.: Deploy redundancy of Internet using first Hop redundancy protocol and
monitoring it using IP service level agreements. Int. J. Eng. Sci. Comput. 7(10), 15320–15322.
Pearl Media Publications, Bangalore (2017)
33. Rahman, Z., et al.: Performance evaluation of first HOP redundancy protocols (HSRP, VRRP
& GLBP). J. Appl. Environ. Biol. Sci. 7(3), 268–278. Textroad, Egypt (2017)
34. Froom, R., Sivasubramanian, B., Frahim, E.: Implementing Cisco Switched Networks. Cisco,
Indianapolis (2010)
35. Arthur, C., et al.: The effects of packet reordering in a wireless multimedia environment. In:
1st International Symposium on Wireless Communication Systems, pp. 453–457 (2004)
36. Cheng, Y., Wu, K., Zhang, Q.: From QoS to QoE: a tutorial on video quality assessment. IEEE
Commun. Surv. Tutor. 17(20), 1126–1165 (2015)
37. Gharai, L., Perkins, C., Lehman, T.: Packet reordering, high speed networks and transport
protocol performance. In: 13th International Conference on Computer Communications and
Networks, 1401591, pp. 73–78 (2004)
38. Video Quality of Service (QOS) Tutorial. https://www.cisco.com/c/en/us/support/docs/quality-
of-service-qos/qos-video/212134-Video-Quality-of-Service-QOS-Tutorial.html. Accessed 21
Jan 2019
39. Understanding—and Reducing—Latency in Video Compression Systems. http://www.
cast-inc.com/blog/white-paper-understanding-and-reducing-latency-in-video-compression-
systems. Accessed 21 Jan 2019
40. IP Networks for Broadcaster Applications. https://www.researchgate.net/publication/
308764506_IP_Networks_for_Broadcaster_Applications. Accessed 21 Jan 2019
41. Tao, S., Apostolopoulos, J., Guerin, R.: Real-time monitoring of video quality in IP networks.
IEEE/ACM Trans. Netw. 16(5), 1052–1065 (2008)
42. Hot Standby Router Protocol (HSRP): Frequently Asked Questions. https://www.cisco.com/c/
en/us/support/docs/ip/hot-standby-router-protocol-hsrp/9281-3.html. Accessed 21 Jan 2019
43. Standby-hold-timer. https://www.ibm.com/support/knowledgecenter/SS9H2Y_7.5.0/com.
ibm.dp.doc/standby-hold-timer_interface.html. Accessed 21 Jan 2019
44. Timers (GLBP). https://www.cisco.com/c/m/en_us/techdoc/dc/reference/cli/n5k/commands/
timers-glbp.html. Accessed 21 Jan 2019
45. VRRP failover-delay Overview. https://www.juniper.net/documentation/en_US/junos/topics/
concept/vrrp-failover-delay-overview.html. Accessed 21 Jan 2019
46. White Paper—One-Way Delay Measurement Techniques. https://www.accedian.com/wp-
content/uploads/2015/05/One-WayDelayMeasurementTechniques-AccedianWhitePaper.pdf.
Accessed 21 Jan 2019
Developing a Method for Measuring the Failover Times … 15

47. Why-micro-seconds-are-not-enough. https://packeth.wordpress.com/2015/05/04/why-micro-

seconds-are-not-enough/. Accessed 21 Jan 2019
48. Srivastava, S. et al.: Comparative study of various traffic generator tools. In: Recent Advances
in Engineering and Computational Sciences (RAECS), pp. 1–6 (2014)
49. Srivastava, S. et al.: Evaluation of traffic generators over a 40 Gbps link. In: Asia-Pacific
Conference on Computer Aided System Engineering (APCASE), pp. 43–47 (2014)
50. Implementing Quality of Service Over Cisco MPLS VPNs. http://www.ciscopress.com/
articles/article.asp?p=471096&seqNum=6. Accessed 21 Jan 2019
Random documents with unrelated
content Scribd suggests to you:
 1403. Cp. Metam. xv. 27.
1413 f. Pont. i. 3. 57 f.
1420. Cp. Her. iii. 24, used here with a change of meaning.
1424. Cp. Ars Amat. ii. 88, ‘Nox oculis pavido venit oborta metu.’
1425 f. Pont. i. 2. 45 f.
1429 f. Cp. Pont. i. 2. 49 f.
1433. Metam. iii. 709.
1442. Cp. Her. v. 14, where we have ‘Mixtaque’ instead of
‘Copula.’
1445 ff. Cp. Metam. xiv. 214-216.
1453. Adapted from Metam. iv. 263, ‘Rore mero lacrimisque suis
ieiunia pavit.’ The change of ‘mero’ to ‘meo’ involves a tasteless
alteration of the sense, while the sound is preserved.
1459. Cp. Rem. Amoris, 581.
1465. Metam. ii. 656. Our author has borrowed the line without
supplying an appropriate context, and the result is nonsense. Ovid
has

‘Suspirat ab imis
Pectoribus, lacrimaeque genis labuntur obortae.’

1467 f. Pont. i. 2. 29 f.
1469. Cp. Metam. xiii. 539.
1473. Ovid, Metam. viii. 469.
1475. Metam. iv. 135, borrowed without much regard to the
context.
1485. From Ovid, Her. xiv. 37, where however we have ‘calor,’
not ‘color,’ a material difference.
1496. Her. v. 46.
 1497. The expression ‘verbis solabar amicis’ is from Ovid (Fasti,
v. 237), but here ‘solabar’ seems to be made passive in sense.
1501 f. i.e. ‘cessat amor eius qui prius,’ &c., with a rather harsh
ellipse of the antecedent. The couplet is a parody of Ovid, Pont. iv. 6.
23 f.,

‘Nam cum praestiteris verum mihi semper

amorem,
Hic tamen adverso tempore crevit amor.’

1503 f. Cp. Tristia, iii. i. 65 f.,

‘Quaerebam fratres, exceptis scilicet illis,

Quos suus optaret non genuisse pater.’

1506. Fasti, i. 148, not very appropriate here.

1512. Her. xi. 82.
1514. Cp. Her. xiii. 86, ‘Substitit auspicii lingua timore mali.’
1517 f. Cp. Her. iii. 43 f.
1519. Cp. Pont. iii. 4. 75.
1521. Cp. Tristia, i. 11. 23.
1534. Cp. Tristia, v. 4. 4, ‘Heu quanto melior sors tua sorte mea
est.’
1535 ff. Cp. Tristia, iii. 3. 39 ff.
1539 f. Tristia, iii. 3. 29 f.
1541. scis quia: ‘quia’ for ‘quod,’ cp. l. 1593; ‘puto quod,’ i. Prol.
15, &c.
1549. Cp. Fasti, i. 483.
1564. Her. xiv. 52.
1565 f. Cp. Her. x. 113 f. The lines are not very appropriate here.
1568. See note on l. 1420.
 1569. Cp. Metam. iii. 396.
1571. Cp. Metam. xiv. 210.
1573. Cp. Metam. vii. 614.
1575. Cp. Metam. ix. 583.
1581. Obice singultu, that is, ‘Impediente singultu’: cp. Cronica
Tripertita, ii. 3.
1585. Metam. xiv. 217.
1589. Tristia, i. 5. 45.
1593. vidi quia: cp. l. 1541.
1609. quid agant alii, ‘whatever others may do.’
1612. Cp. Her. xix. 52.
1615 f. It seems probable that this is a prayer to the Virgin Mary,
whose name ‘Star of the Sea’ was used long before the fourteenth
century, e.g.

‘Praevia stella maris de mundo redde procella

Tutos: succurre, praevia stella maris,’

in an address to the Virgin by Eberhard (date 1212) in Leyser, Poet.

Med. Aevi, p. 834, and the name occurs also in Peter Damian’s
hymns (xi. cent.). For Gower’s use of the expression cp. Mirour de
l’Omme, 29925, ‘O de la mer estoille pure,’ and later in this book, l.
2033, ‘Stella, Maria, maris.’ Here, however, we might translate, ‘Be
thou a star of the sea going before me,’ taking it as a prayer to
Christ.
1623. Metam. i. 265.
1627. Extra se positus, ‘beside himself.’
1630. Fasti, iv. 386.
1631. Cp. Metam. i. 282.
1635. Cp. Metam. i. 269.
 1637. Cp. Metam. i. 270.
1653 ff. From this point to the end of the chapter the description
is mostly taken from Ovid, Metam. xi. 480-523, many hexameters
being appropriated without material change, e.g. ll. 480, 482, 484,
486, 488, 491, 492, 495, 499, 501, 516, 517, 519 f.
1689. The line is taken away from its context, and consequently
gives no sense. In Ovid it is,

‘Ipse pavet, nec se qui sit status ipse fatetur

Scire ratis rector.’—Metam. xi. 492.

1693. Metam. i. 292.

1695. From Peter Riga, Aurora, (MS. Bodley 822) f. 16 vo.
1697-1700. Cp. Aurora, f. 15 vo,

‘Fontes ingresso Noe corrumpuntur abyssi,

Et de uisceribus terra fluenta uomit.
Effundunt nubes pluuias, deciesque quaternis
Sustinet inmensas archa diebus aquas.’

1717 f. Cp. Ovid, Metam. iv. 689 f.

1719. Cp. Metam. iv. 706 f. Ovid has ‘praefixo,’ which is more
satisfactory.
1721. Cp. Metam. iv. 690.
1727 f. Tristia, i. 11. 21 f.
1729. Fasti, iii. 593.
1735. Metam. xi. 539.
1739. Cp. Metam. xi. 515, ‘Rima patet, praebetque viam letalibus
undis.’
1774. Cp. Fasti, ii. 98.
1775 f. Cp. Amores, ii. 11. 9 f.
1779 f. Tristia, v. 12. 5 f.
 1781. Metam. xiv. 213.
1825. Cp. Tristia, ii. 179.
1832. Tristia, i. 5. 36.
1847 f. Cp. Ovid, Pont. iii. 7. 27 f. In the second line Ovid has
‘tumidis,’ for which there is no authority in Gower. Our author
perhaps read ‘timidis’ in his copy of Ovid, or made the change
himself, taking ‘timidis’ to mean ‘fearful.’
1879. ‘Perhaps that day would have been the last of confusion,
even if,’ &c. This, by the context, would seem to be the meaning.
1898. Ovid, Fasti, iv. 542.
1899 f. Cp. Pont. i. 3. 9 f.
1907 f. From Godfrey of Viterbo, Pantheon, p. 82 (ed. 1584).
1909. ‘But he who walked upon the sea,’ &c., that is, Christ.
1913. Cp. Metam. i. 328, ‘Nubila disiecit, nimbisque Aquilone
remotis.’
1917. Metam. i. 329.
1919. Cp. Metam. i. 345.
1921. Cp. Metam. v. 286, where we have ‘nubila,’ as the sense
requires. Here the MSS. give ‘numina’ without variation.
1923. Cp. Metam. ix. 795.
1925. Metam. i. 344.
1935. Metam. xiii. 440.
1939. Metam. xiii. 419.
1944. Quam prius: for ‘prius quam,’ as often.
1963 f. This alludes to the supposed reply made to Brutus (son of
Silvius), when he consulted the oracle of Diana in the island of
Leogecia, ‘Brute, sub occasum solis,’ &c., as told by Geoffrey of
Monmouth.
1979 f. Ovid, Pont. iii. 8. 15 f.
 1991 ff. Cp. Tristia, i. 11. 25 ff.
1997 f. Tristia, iii. 2. 25 f.
2001 f. Cp. Her. xi. 27 f.
2003 f. Her. xiv. 39 f.
2029 f. Cp. Rem. Amoris, 119 f.
2031 f. Cp. Rem. Amoris, 531 f.
2033 f. Cp. Her. ii. 123 f.
2037 f. Pont. iv. 3. 49 f.
2043. Cp. Pont. i. 4. 21. In Ovid we read ‘animus quoque pascitur
illis,’ and this probably was what Gower intended to write.
2071 f. Cp. Pont. ii. 7. 9 f.
2074. Pont. ii. 7. 8.
2091. Cp. Hist. Apollonii Tyrii, xli, ‘Sicut rosa in spinis nescit
compungi mucrone.’
2139. Cp. Pont. i. 5. 47.
2150. Cp. Rem. Amoris, 484.

LIB. II. Prologus.

15. Cp. Speculum Stultorum, p. 11, l. 41 (Rolls Series, 59, vol. i.).
41. Deut. xxxii. 13, ‘ut sugeret mel de petra oleumque de saxo
durissimo.’
49 f. Cp. Fasti, i. 73 f.
51. The supposed mischief-maker is compared to Sinon, who
gave a signal by fire which led to the destruction of Troy: cp. Conf.
Amantis, i. 1172. I cannot satisfactorily explain ‘Excetra.’
57 f. From Neckam, De Vita Monachorum, p. 175 (ed. Wright,
Rolls Series, 59, vol. ii.).
61. De modicis ... modicum: cp. Mirour de l’Omme, 16532.
 64. Cp. Ovid, Ars Amat. ii. 166.

LIB. II.
With the general drift of what follows cp. Conf. Amantis, Prol. 529
ff.
1. Incausti specie, cp. Conf. Amantis, viii. 2212.
18. nos: meaning the people of England, as compared with those
of other countries.
31 f. Cp. Ovid, Tristia, v. 8. 19 f.
33. Tristia, v. 5. 47.
41. Job v. 6, ‘Nihil in terra sine causa fit’: cp. Mirour de l’Omme,
26857.
59. This is the usual opposition of rose and nettle, based perhaps
originally on Ovid, Rem. Amoris, 46: cp. Conf. Amantis, ii. 401 ff.
67 f. Cp. Boethius, Consol. Phil. 2 Pr. 4, ‘in omni adversitate
fortunae infelicissimum genus est infortunii fuisse felicem.’ So Dante,
Inf. v. 121 ff.,

‘Nessun maggior dolore,

Che ricordarsi del tempo felice
Nella miseria.’

117 ff. Cp. Ovid, Her. v. 109 ff. In l. 117 ‘siccis’ is substituted, not
very happily, for ‘suci.’
138. Cp. Conf. Amantis, Latin Verses after ii. 1878,

‘Quod patet esse fides in eo fraus est, que politi

Principium pacti finis habere negat.’

163 f. Cp. Ovid, Tristia, v. 8. 15 f.

167 ff. Cp. Tristia, i. 5. 27 ff.
199 f. There seems to be no grammatical construction here.
 239 ff. With this passage cp. Mirour de l’Omme, 27013 ff., where
nearly the same examples are given. The classification is according
to the nature of the things affected, first the heavenly bodies, then
the elements of air, water, fire and earth, and finally living creatures.
This arrangement is more clearly brought out in the Mirour.
259. Cp. Mirour, 27031, and note.
261. ‘And from the hard rocks of the desert,’ the conjunction
being out of its proper place, as in i. 407, 847, ii. 249, &c.
267 f. Cp. Mirour, 27049 ff.
281 ff. See Mirour de l’Omme, 27073 ff.
282. Congelat, ‘took form.’ Probably the author had in his mind
the phrase ‘congelat aere tacto,’ Ovid, Metam. xv. 415.
306. ‘num’ is here for ‘nonne’; cp. l. 320.
316. Cumque, for ‘Cum’: cp. l. 545, iii. 958, &c.
353 f. Cp. Godfrey of Viterbo, Pantheon, p. 9 (ed. 1584).

‘Ante creaturam genitor deus et genitura,

Primaque natura, novit statuitque futura.’

357-359. These three lines are from the Pantheon, p. 9.

371-374. Taken with slight change from the Pantheon, p. 10.
377 f. From Aurora, (MS. Bodley 822) f. 7 vo.
414. ‘That which the new star brings argues that he is God.’
423. That is, ‘Lux venit, vt obscurari possit tenebris,’ &c.
485. ‘Every one who thinks upon Jesus ought to resolve to lay
aside,’ &c.
487. The MSS. give ‘benedicti,’ but it seems probable that
‘benedici’ was meant. The verb is commonly transitive in later Latin.
495 ff. Cp. Isaiah, xliv. 9-20.
531 f. Psalms, cxiii. 8.
 619 ff. Cp. Ovid, Metam. i. 74 ff.

LIB. III. Prologus.

11 ff. The author characteristically takes care to point out that in
his criticism of the Church he is expressing not his own private
opinion, but the ‘commune dictum,’ the report which went abroad
among the people, and the ‘vox populi’ has for him always a high
authority. Cp. Mirour de l’Omme, 18445 ff., 19057 ff., and see below,
l. 1267 ff, iv. 19 f., 709 f.
With what is said in this Book of the condition of the Church and
the clergy we may compare the author’s Mirour de l’Omme, 18421-
20832.
25 f. Compare with this the author’s note on Mirour de l’Omme,
21266-78.
61. Cp. Ovid, Pont. iv. 14. 41.
64. Cp. Pont. iv. 9. 10.
67 f. Cp. Tristia, ii. 301 f.
82. Cp. Pont. ii. 2. 128.

LIB. III.
1-28. The form of these lines which stood originally in S is given
by the Trinity College, Dublin, and the Hatfield MSS. The passage
has been rewritten over erasure in CHG, and it must be left doubtful
what text they had originally. From the fact that the erasure in G
begins with the second line, it may seem more probable that the
original text of this manuscript agreed with that which we have now
in S, rather than with TH₂: for in the latter case there would have
been no need to begin the erasure before l. 4. In CH the whole
passage has been recopied (the same hand appearing here in the
two MSS.) so that we can draw no conclusion about the point where
divergence actually began. EDL have the same text by first hand. It
will be noted that the lines as given by TH₂ make no mention of the
schism of the Papacy.
11 ff. With this we may compare Mirour de l’Omme, 18769 ff.
22. nisi, for ‘nil nisi’: cp. l. 32.
41. Cp. Ovid, Amores, iii. 8. 55.
63. Fasti, i. 225.
65 f. Cp. Fasti, i. 249 f.
85-90. Chiefly from the Aurora of Petrus (de) Riga, (MS. Bodley
822) f. 71,

‘Ollarum carnes, peponum fercula, porros,

Cepas pro manna turba gulosa petit.
Quosdam consimiles sinus ecclesie modo nutrit,
Qui pro diuinis terrea uana petunt.
. . . . .
Carnes ollarum carnalia facta figurant
Que uelut in nostra carne libido coquit.’

It would seem that Gower read ‘Gebas’ (which has no meaning)

for ‘Cepas’ and ‘preponunt,’ as in MS. Univ. Coll. 143, for ‘peponum,’
which is the true reading, meaning ‘melons’ or ‘pumpkins.’
115. Cp. Metam. xv. 173.
Cap. iii. Heading. Cp. Conf. Amantis, Prol. 288 (margin), where
this is given as a quotation from Gregory.
141 f. Cp. Mirour de l’Omme, 18553.
167 f. From Aurora, f. 37.
175. gregis ex pietate mouetur, ‘is moved by pity for his flock.’
193 ff. With this passage compare Conf. Amantis, Prol. 407-413,
and Mirour de l’Omme, 20161 ff. In all these places a distinct charge
is brought against the clergy, to the effect that they encourage vice,
in order to profit by it themselves in money and in influence: ‘the
prostitute is more profitable to them than the nun,’ as our author
significantly says in the Mirour (20149).
209 ff. Cp. Mirour de l’Omme, 20113 ff.
227 ff. For this attack on the ‘positive law’ of the Church cp. Conf.
Amantis, Prol. 247, Mirour, 18469 ff. The ‘lex positiva’ is that which is
enjoined not as of inherent moral obligation, but as imposed by
Church discipline.
249 f. Cp. Mirour, 18997 ff. Apparently ‘nouo’ is an adverb,
meaning ‘anew,’ ‘again’: cp. 284, 376.
265 ff. Cp. Mirour, 18505 ff.
283 ff. Cp. Mirour, 18637, Conf. Amantis, ii. 3486.
329 ff. With this chapter compare Mirour, 18649-18732.
375. The note which we find here in the margin of SCHGD refers
to the crusade of the bishop of Norwich in Flanders in the year 1383,
which probably took place soon after the completion of our author’s
book. It is added in SCHG in what appears to be one and the same
hand, possibly that of the author himself. If we may judge by the
manner in which the campaign in question is referred to by
contemporary chroniclers, it seems to have been considered a public
scandal by many others besides Gower.
419. Gower uses ‘sublimo’ as an ablat. sing, in l. 701; therefore
‘sublimis’ may here be an ablative plural agreeing with ‘meritis.’
425 ff. Cp. Aurora, (MS. Bodley 822) f. 103,

‘Cogitat inde domum domino fundare, sed audit

A domino, “Templi non fabricator eris.
Es uir sanguineus, ideo templum mihi dignum
Non fabricare potes, filius immo tuus.”
Sanguineus uir signat eum qui, crimina carnis
Amplectens, templum non ualet esse dei.
Ecclesie sancte talis non erigit edem,
Nec sacre fidei collocat ille domum.’
 508. ‘And whosoever may sound trumpets, we ought to be silent’;
cp. i. 1609.
531 f. Aurora, f. 75 vo.
619 f. Ovid, Pont. ii. 5. 61 f.
623 f. Pont. ii. 6. 21 f.
641. See Ars Amat. ii. 417, where we find ‘semine,’ a reading
which is required by the sense, but not given in the Gower MSS.
651. ‘The line of descent by right of his mother proclaims Christ
to be heir of that land in which he was born.’ The author argues for
crusades to recover the Holy Land, if there must be wars, instead of
wars against fellow Christians, waged by one pope against the other
under the name of crusades: cp. below, 945 ff.
676. quo foret ipse vigil, ‘where it ought to be watchful,’ a
common use of the imperfect subjunctive in our author’s Latin: cp.
‘gestaret,’ 695, ‘lederet,’ 922, ‘medicaret,’ 1052.
815. What follows is spoken as in the person of the supreme
pontiff: cp. Mirour, 18505-18792, where somewhat similar avowals
are put into the mouth of a member of the Roman Court.
819 f. Cp. Conf. Amantis, Prol. 261,

‘The hevene is ferr, the world is nyh.’

835. Ovid, Fasti, v. 209.

955 f. I take this concluding couplet as a remark made by the
author on the sentiments which he has just heard expressed by the
representative of the Pope. It practically means that ‘Clemens’ is not
a proper name for the Pope: it is in fact a ‘headless name’ and
should rather be ‘Inclemens.’ Compare the address to Innocent III at
the beginning of Geoffrey de Vinsauf’s Poetria Nova:

‘Papa, stupor mundi, si dixero Papa nocenti,

Acephalum nomen tribuam tibi: si caput addam,
Hostis erit metri,’ &c.
 957 ff. It seems best to take what follows as, in part at least, a
dialogue between the author and the representative of the pope, who
has just spoken. Soon however the speech passes again entirely to
the author. The Biblical reference here is to Revelation, xxii. 8 f. The
same use is made of it in the Mirour, 18736 ff.
1077-1080. These four lines are from the Aurora, f. 44 vo.
1113 f. Ovid, Ars Amat. iii. 595 f. (where we have ‘sequatur’). The
original application is to the effects of rivalry in stimulating the
passion of lovers. For the use of ‘sequetur’ here, apparently as a
subjunctive, compare l. 1946, ‘Inueniet tardam ne sibi lentus opem.’
1118-1124. These lines are almost entirely borrowed from the
Aurora, (MS. Bodley 822) f. 21 vo.
1124. In the Glasgow MS. ‘Est’ has been here altered to ‘Et.’
1145-1150. Almost verbatim from Aurora, f. 93 vo.
1169. S has here in the margin in a somewhat later hand than
that of the text, ‘Nota hic quattuor neccessaria episcopo.’
1171 f. Cp. Aurora, f. 44 vo,

‘Est olei natura triplex, lucet, cybat, unguit;

Hec tria mitratum debet habere capud.’

1183 f. Cp. Aurora, f. 44 vo,

‘Lux est exemplo, cibus est dum pascit egenos,

Vnctio dum populis dulcia uerba ferit.’

Gower is right in reading ‘serit,’ which is given in MS. Univ. Coll. 143,
f. 13.
1206. Cp. l. 1376.
1213. Cp. Ovid, Ars Amat. iii. 655.
1215 f. Cp. Ars Amat. iii. 653 f.
1233. Cp. Ars Amat. ii. 279.
 1247 ff. Cp. Mirour, 18793 ff.
1267. Vox populi, &c.: cp. Speculum Stultorum, p. 100, l. 4, and
see also the note on iii. Prol. 11.
1271. Cp. Conf. Amantis, Prol. 304 ff. and Mirour, 18805.
1313. With the remainder of this Book, treating of the secular
clergy, we may compare Mirour de l’Omme, 20209-20832.
1341. Cp. Mirour, 18889 ff.
1342. participaret, ‘he ought to share’: see note on l. 676.
1359 f. Cp. Conf. Amantis, i. 1258 ff.
1375 ff. Cp. Mirour, 20287 ff.
1376. The reading ‘vngat vt’ is given by the Digby MS. and
seems almost necessary: cp. l. 1206.
1405. prece ruffi ... et albi, ‘by reason of the petition of the red
and the white,’ that is, presumably, by the influence of gold and
silver, ‘dominis’ in the next line being in a loose kind of apposition to
a dative case suggested by ‘Annuit.’
1407. S has here in the margin, in a rather later hand, ‘contra
rectores Oxon.’
1417. Eccles. iv. 10, ‘Vae soli, quia cum ceciderit, non habet
sublevantem se.’
1432. The margin of S has here, in the same hand as at 1407,
‘Nota rectores et studentes Oxon.’
1443. formalis, that is, ‘eminent,’ from ‘forma’ meaning ‘rank’ or
‘dignity,’ but here also opposed to ‘materialis.’
1454. Originally the line was ‘Dum legit, inde magis fit sibi sensus
hebes,’ but this was altered to ‘plus sibi sensus hebes est,’ with the
idea apparently of taking ‘magis’ with ‘legit.’ This involves an
awkward metrical licence, ‘hebes est’ equivalent to ‘hebest,’ and the
original text stands in CEH as well as in TH₂. The expedient of the
Roxburghe editor is quite inexcusable.
 1493 ff. Cp. Mirour, 20314. The sporting parson was quite a
recognized figure in the fourteenth century. Readers of Froissart will
remember how when the capture of Terry in Albigeois was effected
by stratagem, the blowing of the horn to summon the company in
ambush was attributed by those at the gate to a priest going out into
the fields, ‘Ah that is true, it was sir Francis our priest; gladly he
goeth a mornings to seek for an hare.’
1498. fugat: used apparently as subjunctive also in l. 2078, but it
is possible that ‘Nec fugat’ may be the true reading here.
1509 ff. Cp. Mirour de l’Omme, 20313 ff.
1527. Est sibi missa, ‘his mass is over.’
1546. Apparently a proverbial expression used of wasting
valuable things.
1549. If benefices went from father to son, little or nothing would
be gained by those who go to Rome to seek preferment, for an heir
would seldom fail.
1555 ff. Cp. Mirour de l’Omme, 20497 ff. The priests here spoken
of are the ‘annuelers,’ who get their living by singing masses for the
dead, the ‘Annua seruicia’ spoken of below:

‘Et si n’ont autre benefice,

Chantont par auns et par quartiers
Pour la gent mort.’ Mirour, 20499.

1559. In the Mirour,

‘Plus que ne firont quatre ainçois’ (20527).

1587-1590. Taken with slight change from Aurora, (MS. Bodley

822) f. 65 vo.
1591. ‘With the ancients it is possible to say “hic et hec
sacerdos,”’ that is, ‘sacerdos’ is both masculine and feminine.
1693-1700. Adapted from Aurora, f. 65,
 ‘Omen in urbe malum bubo solis iubar odit,
Escam uestigat nocte, ueretur aues:
In quem forte gregis auium si lumina figant,
Et clamando uolant et laniando secant.
Incestus notat iste reos, qui corpore fedi
Contra nature iura latenter agunt:
Hos iusti quasi lucis aues discerpere querunt,
Zelo succensi uerba seuera serunt.’

(‘Conclamando’ for ‘Et clamando’ in MS. Univ. Coll. 143.)

1727 ff. Cp. Mirour, 20713 ff.
1759 ff. Cp. Mirour, 20725 ff.,

‘Ne sont pas un, je suis certeins,

Ly berchiers et ly chapelleins,
Ne leur pecché n’est pas egal,
L’un poise plus et l’autre meinz,’ &c.

1775. fierent, ‘ought to become’: cp. l. 1789.

1791-1794 are from Aurora, f. 93 vo, and the succeeding couplet
is adapted from the same source, where we have,

‘De lino que fit per ephot caro munda notatur,

Nam tales seruos Cristus habere cupit.’

1797. Cp. Aurora, f. 46 vo,

‘Balteus ex bysso tunicam constringit honeste.’

1799 f. Cp. Aurora, f. 45 vo.

1801 f.

‘In medio tunice capitale ligat sibi presul,

Vt capitis sensus non sinat ire uagos.’
Aurora, f. 46.

1807 f.
 ‘Aurum ueste gerit presul, cum splendet in illo
Pre cunctis rutilans clara sophia patris.’
Aurora, f. 45.

1809 ff.

‘Ne tunice leuiter possit ruptura minari,

Illius in gyro texilis ora micat:
A grege ne presul se disrumpat, set honestus
Ad finem mores pertrahat, ista notant.’
Aurora, f. 46.

1813 f. Cp. Aurora, f. 46 vo.

1815-1818.

‘Aaron et natis uestes texuntur, ut horum

Quisque sacerdotis possit honore frui.
Nam modo presbiteri, seu summi siue minores,
Conficiunt Cristi corpus idemque sacrant.’
Aurora, f. 45.

1823 f. Aurora, f. 43 vo.

1841-1848. These eight lines are taken with insignificant changes
from the Aurora, f. 63 vo.
1853. The reference here given by Gower to the Aurora of Petrus
(de) Riga has led to the tracing of a good many passages of the Vox
Clamantis, besides the present one, to that source.
1863-1884. These lines are almost entirely from Aurora, ff. 66 vo,
67. The arrangement of the couplets is somewhat different, and
there are a few slight variations, which are noted below as they
occur.
1866. eius: ‘illud,’ Aurora, f. 67.
1868. tumet: ‘timet,’ Aurora. (MS. Bodley 822), but Gower’s
reading is doubtless the more correct.
1871. nimio: ‘magno,’ Aurora.
 1872. ipse: ‘esse,’ Aurora.
1876. ligante: ‘trahente,’ Aurora, f. 66 vo.
1878. tardat ad omne bonum: ‘ad bona nulla ualet,’ Aurora.
1880. Lumina nec: ‘Nec faciem,’ Aurora, f. 67.
1881 f.

‘Per pinguem scabiem succensa libido notatur;

Feruet vel fetet corpus utroque malo.’
Aurora, f. 67.

1885 ff. Our author still borrows from the same source, though
from a different part of it. We find these lines nearly in the same form
in the Aurora, f. 103,

‘Oza manus tendens accessit ut erigat archam,

Set mox punita est arida facta manus.
Hinc ideo dicunt meruisse necem, quia nocte
Transacta cohitu coniugis usus erat.
Declaratur in hoc quod si pollutus ad aram
Accedas, mortis uulnere dignus eris.’

1891 f.

‘Namque superiectas sordes detergere pure

Nescit nostra manus, si tenet illa lutum.’
Aurora, f. 103.

1905-1908. These two couplets are from Aurora, f. 69 vo, where

however they are separated by four lines not here given.
1911 ff. Cp. Aurora, f. 69 vo,

‘Radices non extirpat rasura pilorum,

Set rasi crescunt fructificantque pyli.
Sic licet expellas omnes de pectore motus,
Non potes hinc penitus cuncta fugare tamen.
Hec de carne trahis, quia semper alit caro pugnans;
 Intus habes cum quo prelia semper agas.’

Gower’s reading ‘pugnam’ in l. 1915 is probably right.

1937. Ovid, Rem. Amoris, 669.
1939. Tristia, iv. 6. 33 f.
1943 f. Rem. Amoris, 89 f.
1945 f. Cp. Rem. Amoris, 115 f.
1946. Inueniet: apparently meant for subjunctive; cp. l. 1114.
1947-1950. Rem. Amoris, 81-84.
1952. Cp. Her. xvii. 190.
1953. Rem. Amoris, 229.
1955 f. Rem. Amoris, 139 f.
1999 f.

‘Cum sale uas mittens in aquas Helyseus, easdem

Sanat, nec remanet gustus amarus aquis.’
Aurora, f. 140.

2001. Aurora, f. 60 vo.

2017-2020. From Aurora, f. 8.
2035-2040. From Aurora, f. 15 vo, but one couplet is omitted, and
so the sense is obscured. After ‘sunt sine felle boni’ (l. 2038), the
original has,

‘Cras canit hinc coruus, hodie canit inde columba;

Hec vox peruersis, congruit illa bonis.
Cras prauum cantant, dum se conuertere tardant,
Set tales tollit sepe suprema dies.’

The meaning is that the bad priests cry ‘Cras,’ like crows, and
encourage men to put off repentance, while the others sing ‘Hodie,’
like doves, the words ‘cras’ and ‘hodie’ being imitations of the notes
of the two birds. The expression ‘Cras primam cantant,’ in l. 2039, is
not intelligible, and probably Gower missed the full sense of the
passage.
2045. ‘sit’ has been altered in S from ‘fit.’
2049 ff. Cp. Mirour de l’Omme, 20785 ff.
2071. Cp. Mirour, 20798.
2078. fugat: cp. l. 1498.
2097 f. Cp. iv. 959 and note.

LIB. IV.
The matter of this book corresponds to that of the Mirour de
l’Omme, ll. 20833-21780.
19 f. Cp. Lib. iii. Prol. 11.
34. ‘dompnus’ or ‘domnus’ was the form of ‘dominus’ which was
properly applied as a title to ecclesiastical dignitaries, and it seems
to have been especially used in monasteries. Ducange quotes John
of Genoa as follows: ‘Domnus et Domna per syncopen proprie
convenit claustralibus; sed Dominus, Domina mundanis.’ Cp. l. 323
of this book and also 327 ff.
57. humeris qui ferre solebat, ‘who used to bear burdens,’ as a
labourer.
87. Cp. Godfrey of Viterbo, Pantheon, p. 74 (ed. 1584).
91. Pantheon, p. 74.
109 f. Cp. Ovid, Fasti, i. 205 f.
111. Ars Amat. ii. 475, but Ovid has ‘cubilia.’
112. Cp. Fasti, iv. 396, ‘Quas tellus nullo sollicitante dabat.’
Gower has not improved the line by his changes.
114. Fasti, iv. 400.
115. Metam. i. 104, but Ovid has of course ‘fraga.’