Fatigue and Fracture of Weldments The

IBESS Approach for the Determination

of the Fatigue Life and Strength of
Weldments by Fracture Mechanics
Analysis Uwe Zerbst
Visit to download the full and correct content document:
https://textbookfull.com/product/fatigue-and-fracture-of-weldments-the-ibess-approac
h-for-the-determination-of-the-fatigue-life-and-strength-of-weldments-by-fracture-mec
hanics-analysis-uwe-zerbst/
More products digital (pdf, epub, mobi) instant
download maybe you interests ...

Fatigue and Fracture of Nanostructured Materials

Pasquale Cavaliere

https://textbookfull.com/product/fatigue-and-fracture-of-
nanostructured-materials-pasquale-cavaliere/

The Welding Engineer s Guide to Fracture and Fatigue

1st Edition Philippa L Moore

https://textbookfull.com/product/the-welding-engineer-s-guide-to-
fracture-and-fatigue-1st-edition-philippa-l-moore/

Proceedings of the 8th International Conference on

Fracture Fatigue and Wear FFW 2020 August 26 27 2020
Magd Abdel Wahab

https://textbookfull.com/product/proceedings-of-the-8th-
international-conference-on-fracture-fatigue-and-wear-
ffw-2020-august-26-27-2020-magd-abdel-wahab/

The Fracture of Brittle Materials: Testing and Analysis

Stephen W. Freiman

https://textbookfull.com/product/the-fracture-of-brittle-
materials-testing-and-analysis-stephen-w-freiman/
The fracture of brittle materials : testing and
analysis Second Edition Freiman

https://textbookfull.com/product/the-fracture-of-brittle-
materials-testing-and-analysis-second-edition-freiman/

The fracture of brittle materials: testing and analysis

Second Edition Freiman

https://textbookfull.com/product/the-fracture-of-brittle-
materials-testing-and-analysis-second-edition-freiman-2/

A Practical Approach to Fracture Mechanics Jorge Luis

González-Velázquez

https://textbookfull.com/product/a-practical-approach-to-
fracture-mechanics-jorge-luis-gonzalez-velazquez/

Fracture Mechanics: Fundamentals and Applications,

Fourth Edition Ted L. Anderson

https://textbookfull.com/product/fracture-mechanics-fundamentals-
and-applications-fourth-edition-ted-l-anderson/

Geologic Fracture Mechanics 1st Edition Richard A.

Schultz

https://textbookfull.com/product/geologic-fracture-mechanics-1st-
edition-richard-a-schultz/
Uwe Zerbst · Mauro Madia · Benjamin Schork
Jonas Hensel · Pawel Kucharczyk
DesireTchoffo Ngoula · Didi Tchuindjang
Julian Bernhard · Carla Beckmann

Fatigue and
Fracture of
Weldments
The IBESS Approach for the
Determination of the Fatigue Life and
Strength of Weldments by Fracture
Mechanics Analysis
Fatigue and Fracture of Weldments
Uwe Zerbst Mauro Madia
•

Benjamin Schork Jonas Hensel

•

Pawel Kucharczyk DesireTchoffo Ngoula

•

Didi Tchuindjang Julian Bernhard

•

Carla Beckmann

Fatigue and Fracture

of Weldments
The IBESS Approach for the Determination
of the Fatigue Life and Strength of Weldments
by Fracture Mechanics Analysis

123
Uwe Zerbst DesireTchoffo Ngoula
Bundesanstalt für Materialforschung Materials Mechanics Group
und -prüfung (BAM) Technische Universität Darmstadt
Berlin, Germany Darmstadt, Hessen, Germany

Mauro Madia Didi Tchuindjang

Bundesanstalt für Materialforschung Saarbrücken, Saarland, Germany
und -prüfung (BAM)
Berlin, Germany and

Benjamin Schork Hamburg University of Technology (TUHH)

Center for Engineering Materials—State Hamburg, Germany
Materials Testing Institute Darmstadt
(MPA), Institute for Materials Julian Bernhard
Technology(IfW) SAM, LBF Darmstadt
Technische Universität Darmstadt Technische Universität Darmstadt
Darmstadt, Hessen, Germany Darmstadt, Hessen, Germany

Jonas Hensel Carla Beckmann

Institute of Joining and Welding IFS Fraunhofer-Institut für Werkstoffmechanik
Braunschweig University of Technology Freiburg im Breisgau, Baden-Württemberg,
Braunschweig, Niedersachsen, Germany Germany

Pawel Kucharczyk
IWT Solutions AG
Aachen, Nordrhein-Westfalen, Germany

ISBN 978-3-030-04072-7 ISBN 978-3-030-04073-4 (eBook)

https://doi.org/10.1007/978-3-030-04073-4

Library of Congress Control Number: 2018960756

© Springer Nature Switzerland AG 2019

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.

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

This book gives a comprehensive and thorough guide to the readers who are lost in
the forest of fatigue of weld. It presents the clear fracture mechanics and material
background of fatigue of weld by starting first from fatigue crack initiation, short
crack and next by discussing long cracks, crack closure, crack growth and
threshold, residual stress, stress concentration, stress intensity factor, J-integral,
multiple cracks, weld geometries and defects, microstructural parameters including
HAZ and cyclic stress–strain behavior. The authors have succeeded in treating all
these mutually interacting important parameters in a unique distinct way of anal-
ysis. The elaborate works performed by the deep scientific knowledge of the
authors should be admired.
This is the timely and excellent publication in the field of weld.

Fukuoka, Japan Yukitaka Murakami

Emeritus Professor
Kyushu University

v
Notice

While the authors believe that the information and guidance given in this book are
correct, all parties making use of it must rely on their own skill and judgement. The
authors cannot assume any liability for loss or damage caused by any error or
omission in the application of the IBESS method. Any and all such liabilities are
disclaimed.
The authors do not give any warranty or guarantee whatsoever that the infor-
mation and guidance given in this book do not infringe the rights of any third party
or can be used for any particular purpose at all. Any person intending to use the
same should satisfy himself as to accuracy and the suitability for the purpose for
which it is intended to be used.

vii
Contents

Fatigue and Fracture of Weldments . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

1 Background Information . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1 Stages of Fatigue Crack Propagation and Lifetime . . . . . . . . . . . . 1
1.2 The Crack Closure Phenomenon . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.3 Crack Arrest . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.4 Describing Fatigue Crack Propagation and Arrest . . . . . . . . . . . . . 8
1.5 Fatigue Life and Fracture Mechanics Triangle . . . . . . . . . . . . . . . 16
1.6 Initial Crack Size for Fracture Mechanics Analysis . . . . . . . . . . . . 19
1.7 Multiple Crack Propagation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
1.8 The Need of Probabilistic Assessment . . . . . . . . . . . . . . . . . . . . . 21
1.9 Factors Affecting Monotonic Fracture and Fatigue Crack
Propagation of Welds . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
2 The IBESS Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
2.1 Basic Philosophy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
2.2 Cyclic Elastic-Plastic Crack Driving Force . . . . . . . . . . . . . . . . . . 43
2.3 IBESS Input Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
2.4 Determination of the Fatigue Strength and Life . . . . . . . . . . . . . . 96
2.5 Preliminary Recommendations for Considering Welding
Residual Stresses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
2.6 Misalignment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
2.7 Application and Validation of the Approach . . . . . . . . . . . . . . . . . 113
3 Summary and Future Perspective . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128
Annexes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
Annex A: Master Curve Analyses for the Statistical Determination
of Monotonic Fracture Resistance of Welds . . . . . . . . . . . . . . . 133
Annex B: Strength Mis-match ηp Factors for J-Integral
Determination . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137

ix
x Contents

Annex C: Determination of the Monotonic Elastic-Plastic Crack

Driving Force Including Strength Mis-match
and Secondary Stresses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 139
Annex D: Roughness Determination Close to the Weld Toe
for Specifying the Secondary Notch Depth k . . . . . . . . . . . . . . 147
Literature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149
Contributors

J. Baumgartner Fraunhofer Institute for Structural Durability and System

Reliability LBF, Darmstadt, Germany
H. Th. Beier Materials Mechanics Group, Technische Universität Darmstadt,
Darmstadt, Germany
W. Fricke Hamburg University of Technology (TUHH), Hamburg, Germany
J. Hohe Fraunhofer-Institut für Werkstoffmechanik, Freiburg, Germany
M. Kaffenberger Center for Engineering Materials—State Materials Testing
Institute Darmstadt (MPA), Institute for Materials Technology (IfW), Technische
Universität Darmstadt, Darmstadt, Germany
S. Münstermann RWTH Aachen University, Aachen, Germany
T. Nitschke-Pagel Institute of Joining and Welding IFS, Technical University
Braunschweig, Braunschweig, Germany
M. Vormwald Materials Mechanics Group, Technische Universität Darmstadt,
Darmstadt, Germany

xi
Abbreviations

ASTM American Society for Testing and Materials

bcc Body-centered cubic (lattice)
BM Base metal
BS The British Standards Institution
CDF Crack driving force
CG Coarse grain (HAZ)
COV Coefficient of variation (ratio of standard deviation r and mean or
expected value l of a statistical distribution)
CP Cap position at a weld (Fig. 23)
CPCA Compression pre-cracking constant amplitude
CPLR Compression pre-cracking load reduction
C(T) Compact tension specimen (fracture mechanics testing)
cyc Cyclic
FAD Failure assessment diagram
FAT FAT class, stress range referring to 2
106 loading cycles
fcc Face-centered cubic (lattice)
FCP Fatigue crack propagation
FG Fine grain (HAZ)
HAZ Heat-affected zone
IIW International Institute of Welding
ISO International Organization for Standardization
L Longitudinal
LC Long crack
MAG Metal active gas (welding)
NASGRO Computer program for fatigue crack propagation, provided by NASA
OM Strength overmatching (rYW > rYB)
p Plasticity-corrected
p Primary
PWHT Post-weld heat treatment
R-curve Crack resistance curve (monotonic and cyclic version)

xiii
xiv Abbreviations

s Secondary
SC Short crack
SE(B) Single notch bending specimen (fracture mechanics testing)
T Transverse
TIG Tungsten inert gas (welding)
UM Strength undermatching (rYW < rYB)
WM Weld metal
Nomenclature

a Crack length (crack depth for surface cracks)

ai Initial crack depth (for fracture mechanics analysis)
af Final crack depth
a0 El Haddad parameter, Eq. (6)
a* Correction term for modified El Haddad’s model, Eqs. (7) and (8)
a/c Crack aspect ratio (changes during crack propagation and
coalescence)
A Elongation at fracture (uniaxial tensile test)
kbk Burgers vector
b,k Parameters for describing a notch geometry (Fig. 74)
B Specimen thickness (fracture mechanics specimen)
c Half crack length at surface (semi-elliptical crack)
C,n Fit parameters of the da/dN–DK curve in the Paris regime
C Gradient for load reduction in standard FCP threshold testing, Fig. 6.
da/dN Fatigue crack propagation rate
E Modulus of elasticity (Young’s modulus)
E0 ¼E for plane stress and E/(1 – m2) for plane strain conditions
f Crack closure function in the NASGRO equation, Eq. (82)
fr Coefficient in re – Rm correlation
f(Lr) Plasticity correction function (monotonic loading)
f(DLr) Plasticity correction function (cyclic loading)
Fi Components of the boundary correction function of K-factor
solutions
h Weld reinforcement, Fig. 68
h Stress triaxiality (ratio of hydrostatic stress to the equivalent stress; in
Fig. 44 to the yield stress of the weld)
H Width or half width of the weld strip (strength mis-match
consideration)
HB Hardness according to Brinell
HV Hardness according to Vickers

xv
xvi Nomenclature

i Counting parameter; i ¼ 1, 2, 3, …
J J-integral (monotonic loading)
J0.2;BL Resistance against stable crack initiation (monotonic loading)
J0.2 Resistance against stable crack initiation (alternative definition)
k Depth of a secondary notch, e.g., an undercut
k Slope of finite life (high cycle) fatigue S-N curve in double
logarithmic scaling, Eq. (107)
K′, n′ Coefficients of the cyclic stress–strain curve (Ramberg–Osgood)
K Stress intensity factor (K-factor)
KJc Monotonic fracture resistance (formally derived from J-integral)
K0 Scale parameter in three-parameter Weibull distribution
Kmat Fracture resistance, monotonic loading (general term)
Kmax Maximum K-factor in a loading cycle
Kmed 50% percentile value of fracture resistance in the Master Curve
approach
Kmin Minimum K-factor in a loading cycle
Kmin Shift parameter in three-parameter Weibull distribution
Kop K-factor at crack opening
Kr K-factor due to residual stresses
Kr Ordinate of the FAD (¼K/Kmat)
KI K-factor for mode I crack opening (normal to the crack faces)
K Mean value of the K-factor (in the loading cycle)
‘ Section width along the weld toe (IBESS approach)
L Weld width, Fig. 77
Lr Ligament yielding parameter (monotonic loading)
m Shape parameter in three-parameter Weibull distribution
m(z,a) Weight function (K-factor determination)
M Strength mis-match ratio (commonly rYW/rYB)
N Number of loading cycles
N Number of specimens in a statistical test set (Figs. 29, 30, 41)
Nc Number of loading cycles up to fracture
p,q Fit parameters of the threshold and fracture regions of the da/dN–DK
curve, Eqs. (4) and (80)
P Probability
Pf Failure probability
Pt Roughness parameter used for specifying the secondary notch depth
k in the present study, Annex D
R Loading ratio (¼rmin/rmax or Kmin/Kmax)
ReL Lower yield strength (materials showing a Lüders’ plateau)
Rp0.2 0.2% proof strength (materials without Lüders’ plateau)
Rm Uniaxial tensile strength
SCF (Elastic) Stress concentration factor (at weld toe surface)
T Plate thickness
Ti Components of the traction vector for determining the J-integral
Nomenclature xvii

Tp Peak temperature during welding (Fig. 21)

T0 Transition temperature in the Master Curve approach
ui Displacement vector components for determining the J-integral
U Crack closure parameter (DKeff/DK)
U Energy dissipated in monotonic fracture mechanics test
ULC Crack closure factor for long cracks, independent of crack depth a
USC Crack closure factor for short cracks, as function of crack
extension Da
V Plasticity correction factor for secondary stresses
W Specimen width or half width (fracture mechanics specimen)
W Strain energy density (determination of the J-integral)
Y Boundary correction function in K-factor solution
Z Elastic follow-up factor
a Weld flank angle, Figs. 66, 67
ag Constraint parameter in the NASGRO equation, Eqs. (83, 84, 88, 89)
Da Crack extension (monotonic and cyclic)
Dd Cyclic crack tip opening displacement
DJ Cyclic J-integral (cyclic loading)
DJe Elastic part of the cyclic J-integral
DK K-factor range (Kmax–Kmin)
DKeff Effective K-factor range (¼ Kmax–Kop)
DKp (Formally) plasticity-corrected DK, Eq. (29)
DKth Fatigue crack propagation threshold
DKth,eff Intrinsic fatigue crack propagation threshold (no crack closure effect)
DKth,LC Fatigue crack propagation threshold in the long crack regime
DKth,op Crack closure component of the fatigue propagation threshold
DKth,SC Fatigue crack propagation threshold in the (physically) short crack
regime
De Strain range (emax–emin)
Deref Cyclic counterpart to eref
Dei,j Components of the strain tensor range
DLr Ligament plasticity factor (cyclic loading)
Dr Stress range (rmax–rmin)
Drapp Applied stress range (refers to cross section without crack)
Drref Cyclic counterpart to rref
Dri,j Components of the stress tensor range
Dre Endurance limit range
DTi Components of the traction vector range (cyclic J-integral
determination)
DW Cyclic strain energy (cyclic J-integral determination)
Dui Components of the displacement vector range
e Strain
ea Strain amplitude (¼ ½ De)
eref Reference strain (reference stress method)
eij Components of the strain tensor
xviii Nomenclature

gp Geometry function in monotonic J-integral testing

l Mean or expected value of statistical distributions
q Weld toe radius, Figs. 64, 65
q Notch tip radius (Fig. 74)
emax Maximum strain in the fatigue cycle
emin Minimum strain in the fatigue cycle
c(R) Coefficient for mean stress correction of re, Eqs. (105) and (106)
C Integration path for J-integral determination, Fig. 46
m Poisson’s ratio
r Stress
r Standard deviation of statistical distributions
ra Stress amplitude (¼ ½ Dr)
rb Bending stress
rbs Bending stress due to misalignment as referred to the membrane
stress rm, Sect. 2.6
re Endurance limit (stress amplitude), defined for N ¼ 107 in IBESS
ri Polynomial coefficients of the through-thickness stress profile,
Eqs. (35) and (36)
rij Components of the stress tensor
rf Flow stress in the original NASGRO equation [¼ ½(rY + Rm)]
rm Membrane stress (tension loading)
rm First invariant of stress tensor (Fig. 44)
rmax Maximum stress in the fatigue cycle
rmin Minimum stress in the fatigue cycle
rn Maximum remote stress (Eq. 31)
rref Reference stress (reference stress method)
r0 Reference yield stress, Sect. 2.2.3.2.6
r0,m Reference yield stress (membrane; tension loading)
r0,b Reference yield stress (bending loading)
r1 First principal stress (refers to mode I loading in the case of a crack)
r(z); r1(z) Through-thickness stress profile
rY Yield strength, general (either ReL or Rp0.2)
r0Y (Stabilized) cyclic yield strength
rYB Yield strength of base metal
rYW Yield strength of weld metal
rp Primary stress
rr Residual stress
rTr Transverse residual stress
rs Secondary stress
Abstract

The acronym IBESS stands for “Integrale Bruchmechanische Ermittlung der

Schwingfestigkeit von Schweißverbindungen” which, translated from German,
means “integral fracture mechanics determination of the fatigue strength of welds.”
The method introduced in this study is the outcome of a German research cluster in
which eight partners were involved. A list of them is found at the end of this study.
The IBESS method is characterized by a number of partially novel aspects and
elements of fracture mechanics applied to the evaluation of fatigue strength of
welds. The most important ones are:

• Determination of fatigue crack propagation for mechanically/physically short

and long cracks.
• Determination of an elastic–plastic crack driving force for the treatment of
mechanically short cracks. To that purpose, an analytical expression for the
cyclic J-integral was developed and validated against finite element results.
• The gradual buildup of the crack closure phenomenon is determined by using
cyclic R-curves which describe the crack size dependency of the fatigue crack
propagation threshold in the physically short crack growth regime.
• A physically meaningful initial crack size is defined for total life consideration.
It is based on a two-criterion approach. Based on a cyclic R-curve analysis, the
crack size at crack arrest is determined as a lower bound. If, however, a pre-
existing crack-like defect is larger than this, its dimensions define the initial
crack size.
• Multiple crack propagations at the weld toe are considered.
• In conjunction with this, the variation of the weld toe geometry is considered in
a stochastic model.
• As a result, both the fatigue limit (defined for 107 loading cycles) and the finite
life (high cycle) fatigue S-N curve are obtained statistically.
• At various analysis steps, parametric equations have been developed which
allow for analytical calculations instead of complete stochastic analyses based
on finite elements which are unrealistic even at present.

xix
xx Abstract

• The method has been validated with a large number of S-N curves including two
materials, three weldment types with two weld geometries, each referring to
different manufacturing technologies and the as-welded and stress-relieved state.
• Although not finally solved, an extended discussion is provided on the issue of
welding residual stresses including their redistribution under cyclic loading.
• A number of simplifications is proposed at lower analyses levels which, how-
ever, partly lack complete validation by now.
Introduction

The basic topic of the present study is the determination of the fatigue strength by
fracture mechanics. Common fracture mechanics-based fatigue considerations, e.g.,
within the frame of damage tolerance considerations, are usually limited to residual
lifetime determination of so-called long cracks which are often defined in con-
junction with nondestructive testing. The extension of this concept to the total
lifetime, as in the S-N curve approach, requires an adequate description of so-called
short crack propagation which cannot be based on the DK concept and must
consider the crack closure phenomenon as well as its gradual buildup at the short
crack stage. The terms long and short cracks will be explained in Sect. 1.1.3.
Further, it has to provide a meaningful definition of the initial crack dimensions
needed in fracture mechanics and a solution for the multiple crack problems at
stress levels higher than the fatigue limit as this is specific for configurations such as
weldments. A fundamental discussion of these aspects beyond its application to
welds is provided by Zerbst et al. (2018b, c).
The idea to apply fracture mechanics to the determination of the fatigue strength of
weldments is anything but new. First attempts date to around 1970 (Maddox 1970)
who wrote in 1974: “It is now widely recognized that flaws will inevitably exist in
welded structures and the old idea of removing all detectable defects must be replaced
by the ‘fitness for purpose’ design philosophy. This makes it necessary to define
reliable methods of assessing the significance of flaws, particularly in the context of
fatigue, …. The most promising approach to this problem lies in the use of the fracture
mechanics based description of fatigue crack propagation.” (Maddox 1974). At that
time, a major problem was to develop appropriate K-factor solutions. Ever since effort
was spent to further develop the approach. No detailed overview shall be given here;
see, however, Radaj et al. (2006), Chap. 6, for a review. Note that even today, fracture
mechanics besides some promising applications (see, e.g., Nykänen 2005, repre-
sentative for many other studies) is still not widely used as a tool for fatigue strength
determination of weldments. One reason for this is certainly that the approaches
usually fail to meet the above-formulated requirements. They are based on linear
elastic fracture mechanics even for short cracks and on fixed or (from S-N curves)

xxi
xxii Introduction

back-calculated initial crack sizes, they usually consider one crack only, neglect the
variation in the local weld geometry along the weld toe, etc.
The present study aims at a discussion of all these points and at offering possible
solutions obtained within the frame of the German research cluster IBESS. It
combines rather basic notes with the introduction of a methodology for the fracture
mechanics-based determination of the fatigue strength of welds, but also with a
critical discussion of hitherto not satisfactorily resolved problems.
Partners Involved in the IBESS Cluster Project

• Bundesanstalt für Materialforschung und -prüfung Berlin, Division 9.1 (Project

coordination).
• Rheinisch-Westfälische Technische Hochschule (RWTH) Aachen, Institut für
Eisenhüttenkunde (IEHK).
• Technische Universität Braunschweig, Institut für Füge- und Schweißtechnik
(IFS).
• Technische Universität Darmstadt, Institut für Stahlbau und Werkstoffmechanik
(IfSW), Fachgebiet Werkstoffmechanik.
• Technische Universität Darmstadt, Fachgebiet Systemzuverlässigkeit und
Maschinenakustik (SzM), Fraunhofer-Institut für Betriebsfestigkeit und
Systemzuverlässigkeit (LBF).
• Technische Universität, Staatliche Materialprüfungsanstalt Darmstadt und
Fachgebiet und Institut für Werkstoffkunde (MPA/IfW).
• Fraunhofer-Institut für Werkstoffmechanik (IWM) Freiburg.
• Technische Universität Hamburg-Harburg, Institut für Konstruktion und
Festigkeit von Schiffen (SKF).
The authors wish to thank Prof. Dr. Isabel Hadley (The Welding Institute (TWI),
Great Abington, Cambridge) for her valuable discussion of the content of this book.
The authors wish also to thank the Deutsche Forschungsgemeinschaft (DFG)
for the financial funding of four IBESS subprojects. Furthermore, the authors
gratefully appreciate the funding of the four other subprojects by the Arbeitsge-
meinschaft industrieller Forschungsvereinigungen (AiF) network, in particular

xxiii
xxiv Partners Involved in the IBESS Cluster Project

Forschungsvereinigung Stahlanwendung (FOSTA), Forschungsvereinigung

Schweißen und verwandte Verfahren (DVS), Forschungsvereinigung Automo-
biltechnik (FAT), and Center of Maritime Technologies (CMT).
Fatigue and Fracture of Weldments

1 Background Information

1.1 Stages of Fatigue Crack Propagation and Lifetime

1.1.1 Crack Initiation

Fatigue cracks are initiated at the smooth surface in plain specimens (i.e., at surface
extrusions and intrusions caused by persistent slip bands) or—more frequently—at
geometrical discontinuities (which act as geometric stress concentrators) and/or
material defects such as non-metallic inclusions (which cause strain concentration
zones due to their different stiffness compared to the matrix material and sometimes
also chemical mis-match), see Zerbst et al. (2018b). Polak (2003) notes with respect
to the crack nucleation stage: “Numerous studies have shown that in the majority
of materials and under normal loading conditions, the period of crack initiation in
smooth specimens without defects amounts to less than 5–20% of the fatigue life. In
materials containing defects, the fraction of life spent in crack initiation is even lower.
The major part of the life is spent in the growth of cracks, namely in the growth of
short cracks.” That this statement usually is also true with respect to welds has been
confirmed, e.g., by Verreman and Nie (1996) for manual fillet welds of structural
steel.
Subsequent to crack initiation, fatigue crack propagation is characterized by dis-
tinct successive stages, Fig. 1, see, e.g., Tanaka (2003), Davidson et al. (2003).

1.1.2 Microstructurally Short Cracks

As long as its size is still in the order of microstructural features such as the grain
size, the crack is designated as microstructurally short. The crack and the plastic
zone ahead of its tip is completely embedded within a single or a few grains. As the
© Springer Nature Switzerland AG 2019 1
U. Zerbst et al., Fatigue and Fracture of Weldments,
https://doi.org/10.1007/978-3-030-04073-4_1
2 Fatigue and Fracture of Weldments

Fig. 1 Stages of fatigue

crack propagation, schematic
view, according to Miller
(1993), slightly modified

consequence, crack growth is strongly affected by the local microstructure and char-
acterized by (temporary) acceleration and deceleration phases or even crack arrest.
Typical growth barriers are grain boundaries between adjacent grains of different
crystal orientation where the generation of the plastic zone is inferred. An example
is shown in Fig. 2. The microstructural crack propagation stage is terminated when the
crack depth (of the initially semi-circular or semi-elliptical crack) reaches the order
of one or a few grain sizes (Murakami 2002); a few, since not every grain boundary
will act as a crack stopper when, e.g., the crystal orientation of two adjacent grains
is not significantly different. Note that, in general terms, it is not the crack depth
but the crack front length which is the critical parameter. E.g., Capelli et al. (2008),
investigating single phase aluminium alloy, restrict the microstructurally short crack
stage to a crack front length intersecting 15 grains. That also means that the above
statement about the critical crack depth no longer applies for through surface cracks
or long shallow surface cracks showing a small crack aspect ratio a/c. The applica-
tion of classical fracture mechanics to microstructurally short cracks is not possible
(Pippan and Riemelmoser 2003).

1.1.3 Mechanically and Physically Short Cracks

A microstructurally short crack which is not arrested will finally become a mechan-
ically short crack. The general definition of this is a size up to the order of the plastic
zone size ahead of its tip. As the consequence, the linear elastic fracture mechanics
concept based on the K-factor range K is not applicable but elastic-plastic concepts
based on the cyclic J-integral, J, or a cyclic crack tip opening displacement, δ,
have to be used. This will be discussed in Sect. 2.2.
The term “physically short crack” refers to the observation that during that stage
the crack closure phenomenon is being gradually built-up. It does not exist at the very
beginning, but then the effect intensifies with increasing crack size (and crack wake
1 Background Information 3

Fig. 2 Non-propagating crack in EA1N railway axle steel. The crack stopped within a grain. It
experienced more than 108 loading cycles; according to Beretta et al. (2009)

length) and finally it reaches a stable, nearly crack size independent state when the
crack becomes a long one. The crack closure phenomenon is the topic of Sect. 1.2.
Note that the difference in the terms “mechanically” and “physically” short cracks
is not different crack sizes (these at least will overlap) but the different phenomena.

1.1.4 Long Cracks

A mechanically/physically short crack becomes a long one when the crack closure
effect stabilizes, i.e. it becomes independent of the crack size and when the linear
elastic K concept is applicable (see above). An exception is low cycle fatigue (LCF)
which occurs at stress levels close to the yield strength. Excluding this exception,
long crack propagation can be described by the da/dN–K curve and the long fatigue
crack propagation (FCP) threshold Kth,LC . The long crack stage is terminated by
the failure of the structure, i.e., by monotonic fracture but there might also be other
failure criteria associated with loss of functionality.
Figure 3 provides a scheme of the crack sizes referring to the stages discussed
above. Note, however, that the numbers are strongly material dependent. Therefore,
the figure should be regarded as a first impression only which roughly describes the
relations for steel.
4 Fatigue and Fracture of Weldments

Fig. 3 Crack length scales of the life cycle of a component subjected to cyclic loading; according
to Tanaka (2003)

Excursus: A Competing Definition

To avoid confusion, it has to be mentioned that a competing definition of the fatigue
crack propagation stages is also in use, preferably in a practical context. A crack
initiation stage is defined which is not terminated until the crack has reached a size
detectable with realistic effort and a crack propagation stage beyond that crack size.
For steels, the latter roughly refers to the long crack stage in Fig. 3.
Note that the frequent statement, that the fatigue life of welds is controlled by crack
propagation rather than by crack initiation, originally referred to this nomenclature.
In the past it provided the justification to apply long crack fracture mechanics to the
fatigue life determination of welds. Although the initial crack size is reduced to a
small fraction of a millimeter in more recent documents such as the IIW Guidelines
(Hobbacher 2016) long crack fracture mechanics based on the da/dN–K curve is
still in use. However, this is no longer consistent in view of the information provided
in Sect. 1.1.3. As in many other applications, the major part of the fatigue life of
welds is used to be spent at the short crack stage as long as no large initial defects
such as extended lack of penetration exist. Note, however, that the assessment of
those defects is a task for damage tolerance rather than for safe life considerations
based on S-N curve data.

1.1.5 Failure

Failure occurs when the crack reaches a critical size such that the maximum crack
driving force in the loading cycle, e.g. Kmax , reaches or exceeds the monotonic
fracture resistance of the material. Besides this, other failure criteria are possible as
well when the functionality of the component is impaired even if the component does
not break yet.
1 Background Information 5

1.2 The Crack Closure Phenomenon

1.2.1 Brief Overview

Since it can be assumed that a fatigue crack does not grow when the crack is closed,
i.e., its faces are in contact during the lower part of the loading cycle at negative
R ratios (R  Kmin /Kmax ), only part of the stress intensity factor range K (= Kmax
− Kmin ) will contribute to its propagation. This is designated by an effective value
Keff (= Kmax − Kop ) with Kop being the K-factor at crack opening, for the definition
see Fig. 4a. As a first approximation, Keff should be ½ K for R  −1, however,
it is observed to be even smaller, i.e., the crack is open only at stress levels above
zero.1 This is designated as “crack closure phenomenon”. It is commonly described
by a crack closure parameter U such that

Keff  U · K (1)

Fig. 4 Mechanisms of the crack closure phenomenon; a nomenclature; b plasticity-induced mech-

anism; c roughness-induced mechanism; d oxide debris-induced mechanism

1 Sometimes also below, depending on the R-ratio and the applied stress ranges
6 Fatigue and Fracture of Weldments

Several mechanisms are responsible for the crack closure phenomenon. The three
most important ones are the plasticity-induced, the roughness-induced and the oxide
debris-induced crack closure mechanisms (Fig. 4b–d). For these and further ones
see, e.g., Tanaka and Akiniwa (2003) and Suresh (2003).

1.2.2 Plasticity-Induced Crack Closure

Under load, a plastic zone is formed ahead of the crack tip. When the crack prop-
agates, this will remain as plastically stretched material at the surfaces of the wake
of the crack (Fig. 4b). The consequence is geometrical mis-match particularly at the
specimen sides where the plastic zone dimension is largest due to prevailing plane
stress conditions. Since the plastic zone size increases quadratically with the applied
load and decreases quadratically with the yield strength of the material (for small
scale yielding conditions) the plasticity-induced mechanism becomes stronger for
higher maximum loads in the loading cycle and for smaller yield strengths. This is
at least the case for long cracks.

1.2.3 Roughness-Induced Crack Closure

The roughness of the crack faces causes asperity contact (Fig. 4c) which is enhanced
by mixed mode loading components (shear) due to the kinking or branching of the
crack. Usually it will be affected by the grain size in that a larger grain size tends to
a rougher surface. Note that the roughness-induced effect plays a major role in the
threshold regime of the da/dN–K curve.

1.2.4 Oxide Debris-Induced Crack Closure

This mechanism is active for corrosion-prone materials or material conditions. The

crack wake surfaces are covered by oxide layers which thicken due to fretting of
the crack faces at low R ratios. The locally furbished surfaces are again exposed
to corrosion (Fig. 4d). This way a debris layer is gradually formed the thickness of
which is in the order of the crack tip opening displacement in the threshold regime
of the da/dN–K curve.
All crack closure mechanisms become stronger for low and disappear at high R
ratios.
1 Background Information 7

1.3 Crack Arrest

1.3.1 Crack Arrest of Microstructurally Short Cracks

The authors have already pointed to the importance of crack arrest for fatigue life
and strength considerations. It is immediately evident that only cracks which do not
arrest, negatively affect structural integrity. Crack arrest can happen at any of the
crack propagation stages.
As mentioned above, microstructurally short cracks arrest at microstructural bar-
riers, usually grain boundaries. It is important to understand that the fatigue limit is
not a critical stress for crack initiation but a critical stress for avoiding or overcoming
crack arrest (Miller 1993; Murakami 2002). With respect to the material fatigue limit
(determined with smooth, unnotched specimens) it is that stress below which none
of the numerous microstructural short cracks will grow permanently. Note, however,
that in real (usually notched) structures other arrest mechanisms will dominate as
will be discussed below. Note further that the material fatigue limit will disappear
when mechanisms such as corrosion as a time dependent phenomenon are active
which will overcome the crack arrest barriers.

1.3.2 Crack Arrest of Physically Short Cracks

Crack arrest of physically short cracks occurs due to the gradual build-up of the
crack closure effects when the increase of K with crack depth is smaller than the
decrease of Keff due to an increasing Kop .

1.3.3 Crack Arrest of Long Cracks

Long cracks arrest when the applied K is below the fatigue crack propagation
threshold Kth .

1.3.4 Crack Arrest of Short Cracks at Notches

This is probably the most common crack arrest scenario in components (Murakami
2002, 2003; Tanaka 2003). Due to the stress gradient away from the notch the rate
of increase in K becomes smaller, an effect which is additionally superimposed
by the gradual build-up of the crack closure phenomenon. When the resultant Keff
falls below the fatigue crack propagation threshold Kth , crack arrest will occur.
8 Fatigue and Fracture of Weldments

1.3.5 Further Crack Arrest Scenarios

Further crack arrest cases are possible. Three options are: (i) Crack arrest can occur
when the crack grows into a stress shielded area, e.g., near a geometrical stiffener
as it is typical for thin-wall structures. (ii) Arrest also occurs when the crack grows
towards an area of compressive residual stresses (e.g., in deep-rolled crankshafts)
(iii) It also can happen when the crack propagates into a material with a higher
threshold Kth than that from where it started. An example is crack transition from
a fine to a coarse grain zone, e.g. a heat affected zone in a weld. Note that the coarser
grain is usually associated with a higher threshold (Plekhov et al. 2011). A plausible
explanation for this is a stronger roughness-induced crack closure effect when the
larger grain is associated with rougher crack faces, see Sect. 1.2.3.
Whilst, as mentioned above, the crack arrest mechanism of the microstructurally
short cracks (Sect. 1.3.1) determines the material’s fatigue limit, the component
fatigue limit is usually controlled by the other mechanisms.

1.4 Describing Fatigue Crack Propagation and Arrest

1.4.1 Long Fatigue Crack Propagation (da/dN–K Curve)

Long fatigue crack propagation is described by the da/dN–K curve which is a

sigmoidal curve in a log-log plot, Fig. 5.

Fig. 5 Crack growth diagram da/dN–K for long fatigue crack propagation; schematic view
Another random document with
no related content on Scribd:
 Family 1. Syllidae see p. 306 Family 8. Amphinomidae see p. 318
" 2. Hesionidae " 308 " 9. Eunicidae " 318
" 3. Aphroditidae " 309 " 10. Glyceridae " 320
" 4. Phyllodocidae " 313 " 11. Sphaerodoridae " 320
" 5. Tomopteridae " 315 " 12. Ariciidae " 321
" 6. Nereidae " 315 " 13. Typhloscolecidae " 321
" 7. Nephthydidae " 317

Sub-Order 2. Spioniformia.
Family 1. Spionidae see p. 321 Family 4. Magelonidae see p. 325
" 2. Polydoridae " 323 " 5. Ammocharidae " 325
" 3. Chaetopteridae " 323

Sub-Order 3. Terebelliformia.
Family 1. Cirratulidae see p. 325 Family 3. Ampharetidae see p. 330
" 2. Terebellidae " 327 " 4. Amphictenidae " 330

Sub-Order 4. Capitelliformia.
Family. Capitellidae, see p. 331.

Sub-Order 5. Scoleciformia.
Family 1. Opheliidae see p. 331 Family 4. Scalibregmidae see p. 334
" 2. Maldanidae " 332 " 5. Chlorhaemidae " 334
" 3. Arenicolidae " 333 " 6. Sternaspidae " 335

Branch B. Cryptocephala.
Sub-Order 1. Sabelliformia.
Family 1. Sabellidae see p. 336 Family 3. Amphicorinidae see p. 339
" 2. Eriographidae " 338 " 4. Serpulidae " 339

Sub-Order 2. Hermelliformia.
Family. Hermellidae, see p. 341.

Comparative Anatomy of the Polychaeta.

General Shape of the Body.—The majority of the Polychaeta have an elongated

and very mobile body, like that of Nereis, consisting of an indefinite and usually of
a considerable number of segments; a few, however, have a shorter body, with
fewer segments, definite in number, for instance Aphrodite and Polynoë, which
have thirty to forty segments; and some Hesionids, with only some seventeen to
twenty segments.

In Aphroditidae and certain Amphinomidae the body is more or less oval in shape.
In Lipobranchius and Sternaspis it is grub-like, short, and cylindrical, with rounded
ends; in the former it is difficult to distinguish head and tail, or dorsal and ventral
surfaces.

The segments composing the trunk may be all alike, or may constitute two more or
less sharply marked regions, the thorax and abdomen, differing in the character of
the chaetae, or in their arrangement, or in some other way, as in the Sabelliformia
and the Capitelliformia.

As peculiar cuticular structures, the curious shields of Sternaspis, and of certain of

the Maldanidae may be mentioned.

The posterior extremity is generally more or less narrowed, and most of the
Nereidiformia are provided with special elongated cirri, borne by the anal segment.
In the Maldanidae and others the body terminates in a funnel, at the bottom of
which is placed the anus. Only in a few cases is the anus not terminal; in
Notopygos and other Amphinomidae, as well as in some species of Polynoë, it is
dorsal.[313] In Sabellaria and Pectinaria the hinder end of the body undergoes
great degeneration; in the former it is achaetous, but cylindrical and bent forwards
alongside the body (Fig. 131). In Pectinaria (Fig. 177), this region, which is called
the "scapha," is leaf-like, and serves to close the narrower end of the tube in which
the worm lives. Arenicola marina, and some Terebellids have no chaetae in the
hinder, narrower part of the body.

Fig. 131.—Sabellaria alveolata L. × 3. (After Malmgren.) a, Anus.

The Head.—The prostomium is, in the majority of cases, rounded or conical,

though it may be square (Nephthys) or elongated and jointed (Glycera), or even
hammer-shaped (Tomopteris); or it may be fused with the peristomium, and
apparently absent (Arenicola). In the great group Cryptocephala, the peristomium
grows forwards so as to hide the prostomium entirely.

In a few of the Nereidiformia the prostomium is compressed, and in the

Amphinomidae it is provided with a dorsal ridge or "caruncle," which is a leaf-like
process overlapping three or more segments. In many Aphroditidae (as well as in
Polydora) there is a peculiar "frontal" ridge passing forwards from the prostomial
tentacle, and downwards into the mouth (Figs. 132, c, and 133, A, x).

Fig. 132.—Aphrodite aculeata L. Ventral view of anterior region, × 6. a, Prostomium;

c, frontal ridge on prostomium; d, neuropodial cirrus; l, lower lip; m, mouth; p,
palp; s, intersegmental groove; t, tentacle; I, foot of peristomium, which has
shifted forwards so as to lie in front of the mouth; II to V, successive feet.

In all the Nereidiformia, as well as in Sabelliformia and Chlorhaemidae, the

prostomium bears sensory processes of two kinds, viz. dorsal tentacles and
ventral palps. The latter are invariably two in number, and are particularly well
developed in Aphroditidae, Nereidae, Syllidae, some of the Eunicidae, and in
Chlorhaemidae. Even when they are apparently absent, as in Nephthys, it is
possible that they are represented by certain lobes at the sides of the mouth, for in
many Syllidae they are so fused with the prostomium as to be scarcely
distinguishable. In the Chlorhaemids the palps[314] are grooved, and in the
Sabelliformia they become considerably branched, and extend round the
prostomium so as to nearly meet dorsally and ventrally. Each palp is, in this sub-
Order, represented by a greater or smaller number of long, mobile filaments,
arising from a common base; they are grooved along the inner side, ciliated, and
provided with secondary processes. The crown of "gills," in fact, is nothing more
than the greatly subdivided and enormously elongated palps, as both Pruvot[315]
and Meyer[316] have shown. In such forms as Haplobranchus and Amphicorine
the process of subdivision (branching) has only gone a short way. In all the
Sabelliformia each filament, in addition to its sensory function, aids in conveying
food to the mouth by the action of the cilia, and has a blood-vessel within, thus
acting as a respiratory organ. The filament may carry compound eyes (Fig. 143)
either at its apex (Branchiomma) or at intervals along its course (Dasychone).
 Fig. 133.—A, Anterior end of Polydora enlarged. a, Prostomium; x, frontal ridge; I,
peristomium; c', its long cirrus; II, III, etc., the following segments; c, gill; B,
head of Sabellid; P, palps (branchial crown); t, position of tentacles; l,
processes of upper lip membrane; I, peristomium raised into a collar; II, III, IV,
following segments.

In the family Serpulidae one (rarely two) of the most dorsally placed gill filaments
is enlarged terminally, and acts as a stopper or "operculum," which closes the
mouth of the tube when the animal withdraws into it. Further, in Spirorbis this
operculum is grooved on one side, and serves as a brood pouch in which the eggs
undergo development (Fig. 184, p. 341). It will be seen, therefore, that the palps
may be very important organs for the life of the worm, and they are no less
interesting to the comparative anatomist, serving as they do as an excellent
illustration of the various uses which Nature finds for one and the same organ.

In the other sub-Orders the prostomium carries neither palps nor tentacles.

Fig. 134.—Heads of various Polychaeta (diagrammatic). A, Polynoid; B, Syllid; C,

Nephthys; D, Eunice; E, Phyllodoce; F, Trophonia: a, prostomium; c, normal
cirrus; c1, peristomial cirri; c2, cirrus of second segment; c3, cirrus of third
segment; el1 point of attachment of elytron; p, palp: s, nuchal organ (ciliated
pit); t, tentacle; I, peristomium; II, III, IV, segments.

The tentacles in the Nereidiformia present a wide variation in number; probably

the typical number is three, one of which is median and two lateral—as in
Polynoids, Syllidae, and some Eunicidae. Further, there is a certain amount of
evidence in the nerve supply of the median tentacle to show that it was originally
double. The presence of four tentacles, then, as in Nephthys, Phyllodoce, and
Glycera, may be a primitive condition. By the disappearance of the paired lateral
tentacles the worm possesses a single median one, as in Aphrodite and
Amphinomids;[317] whilst a duplication of these lateral ones leads to the condition
of Eunice and Hyalinoecia, which have five tentacles. In the Chlorhaemidae the
number is further increased to five or more on each side,[318] and in the
Terebellidae these prostomial processes become very numerous.

In the Cryptocephala there is never more than a single pair of tentacles, and these
are generally reduced to a group of sensory cells, though in Sabellaria they retain
a considerable size.

In a few genera, such as Aphrodite, Nephthys, Capitella, the first postoral segment
is distinguished from the succeeding segments only by its position with regard to
the mouth (Fig. 132) and by its smaller size. But in the remainder of the
Polychaeta, with here and there an exception, the peristomium is achaetous in the
adult.[319]

Except in the Nereidiformia, peristomial or tentacular cirri are rare, being

represented in the Spioniformia by the very long "tentacles." In the Nereidiformia
one or more of the following segments may be added to the peristomium, and
share in the "cephalisation," which is so characteristic a feature in this group. In
Amphinomids the first three or four chaetigerous segments are incomplete
ventrally, owing to the shifting of the mouth backwards; these segments form
lateral lips, but they are not otherwise modified. In Phyllodoce, however, there are
four cirri on each side of the mouth, and from the arrangement in the Alciopids we
are justified in concluding that the segment which carries the four pairs of cirri is
really made up of three segments (Fig. 134, E). Among the Hesionids there are
four such "cephalised" achaetous segments with long cirri.

Fig. 135.—Sabellaria alveolata L. Ventral view of anterior region, × 10. a, Notopodial

cirrus; b, notopodium; c, neuropodium; ch, peristomial chaetae; d, neuropodial
cirrus; m, mouth; P, multifid palp (gill filaments); P', ridges after removal of gill
 filaments; s, ventral (tubiparous) gland shield; T, tentacle; I, hood formed by
peristomium; II to VI, following segments.

In a few cases, such as the Chlorhaemids and Sternaspis, and to a slight degree
in Arenicola, the "head" and even the anterior part of the worm is capable of being
withdrawn into the body.

The Parapodia and Chaetae.—The typical parts of a parapodium have been

described in the preceding chapter; here it is only necessary to refer to the series
of diagrams (Figs. 136, 137) representing the parapodia of the more common
Polychaetes, and to add a few remarks about them.

Fig. 136.—Parapodia. A, Nephthys; B, Amphinome; C, Glycera (the unlettered lobe

above g is the notopodial cirrus); D, Syllis; E, Eunice; F, Phyllodoce. a,
Notopodial cirrus; b, notopodium; c, neuropodium; d, neuropodial cirrus; g,
special gill; n, aciculum (omitted in B); x, cirriform lip of chaetigerous sac.

In most Annelids the chaetae are in two bundles on each side, but there are
certain families in which the dorsal bundle, and even the notopodium itself, is
absent, as in the Eunicidae, Syllidae, and Phyllodocidae; or the dorsal bundle may
be absent only in certain regions of the body, as in the hind-body of Terebellids. In
some Amphinomidae and Aphroditidae the notopodium is scarcely distinct as a
separate lobe, being a slight tubercle on the upper surface of the neuropodium;
but the notopodial chaetae are present, and indeed particularly well developed in
many cases.

But whilst, in the Nereidiformia, the parapodia, whether consisting of two lobes or
only one, are always well developed, and project to a more or less pronounced
degree from the sides of the body, it is otherwise in the rest of the group, where
the chaetigerous lobes are usually reduced to mere tubercles or ridges, no doubt
in relation to their burrowing or tubicolous habits. In Sternaspis the chaetae issue
directly from the body-wall.
Amongst the Nereidiformia we find examples in which the parapodia, instead of
being more or less conical "legs," are flattened fore and aft so as to serve as
efficient "fins," as in the active swimmers, Nereis virens and Nephthys caeca, and
in the pelagic Phyllodocids, Alciopids, Typhloscolecids, and Tomopteris.

Fig. 137.—Parapodia. A, Polynoë; B, Scoloplos; C, Euphrosyne. (Transverse

section of body.) a', Accessory cirrus; y, doubtful branchiae; D, Sabella
(thoracic). a, Notopodial cirrus ("elytron" in A, "gill" in B); b, notopodium; c,
neuropodium; d, neuropodial cirrus; n, aciculum (accidentally omitted in C).

Of the typical dorsal and ventral cirri, the ventral is only absent in some
Amphinomids amongst the Nereidiformia; the dorsal is absent in Nephthys and
degenerate in Glycera, whilst in a very large number of families of the other sub-
Orders neither cirrus is present. These cirri, though originally filamentous and
sensory, may, by virtue of special blood supply, become "gills," and this occurs in
several families of different sub-Orders. Thus in Eunice this gill is comb-like; in
Amphinome and in Arenicola (on certain segments) it is arborescent, as it is also
in one to three segments in Terebellids; whilst in Ariciidae, Spioniformia,
Cirratulidae, Opheliidae, and Sabellaria it remains more or less finger-shaped or
filamentous. In the family Serpulidae the thoracic cirri, both dorsal and ventral,
become flattened and extended antero-posteriorly, and unite with one another to
form the "thoracic membrane."[320] In Phyllodocidae the cirri are foliaceous and
natatory, and they contain a great quantity of glands of a peculiar character. The
Aphroditidae are distinguished from other Annelids by the possession of "elytra" or
dorsal scales, which appear to be the dorso-ventrally flattened cirri, retaining their
sensory nature, but adding to this function several others.[321]

The chaetae or bristles are mainly used in locomotion, but it is not unreasonable
to believe that some of the stronger, serrated kinds may be used as weapons of
offence and defence; certainly the Polynoids, bristling as they do with stiff chaetae
along each side, must be rather unpleasant to their smaller enemies.

The various bristles may be placed in three chief groups, viz. (1) simple; (2)
jointed; (3) uncini (see Fig. 138).
(1) The simple chaetae may be smooth and hair-shaped, i.e. "capillary," such as
are present in nearly all families: or they may be forked (Amphinomidae), comb-
shaped (Eunice), notched or serrated, or provided with a series of frills at right
angles to their length, as in Aphroditidae; or fringed along one or both sides with a
membranous expansion, as in Terebellids and Sabellids. The simple chaetae may
also be short and spine-like, as in the ventral bundles of Arenicola; or they may be
slightly curved at the end and notched, forming what are generally termed
"crotchets," such as are common amongst Oligochaeta. These "crotchets" may be
simple, or have numerous denticulations at the end (Maldanidae), or be provided
with a membranous hood (Spioniformia, Capitelliformia). In Hermione peculiar
sheathed, spear-like bristles occur (Fig. 138, N).

(2) Jointed chaetae have already been described (p. 246); they are confined to the
sub-Order Nereidiformia, and occur only in certain families.

(3) The uncini are very short chaetae, which are simply embedded in the skin, and
do not extend beyond the body-wall into the body-cavity. An uncinus is a sharply
curved hook, which may have more or less numerous secondary teeth on it. They
are characteristic of the Sabelliformia and the Terebelliformia.

The chaetae appear as solid, usually fibrillated structures, of a yellow or golden

tint, transparent and refringent. Chemically they consist of chitin, and each chaeta
is the product of a single cell. The chaetae of Euphrosyne are hollow and
calcareous, being peculiar in both characters.
 Fig. 138.—Chaetae of various Polychaetes (the magnification is not the same in all
cases). A, Doubly-fringed capillary, from Terebellid; B, hooded crotchet, from
Polydora; C, a fork, from Euphrosyne; D, jointed chaeta, from Phyllodoce; E,
simple chaeta, with serrated ridges or frills, from a Polynoid; F, jointed chaeta,
from Eunice; G, uncinus, from Pomatoceros (Serpulid); H, one of the outer
series of paleae from the hood of Sabellaria spinulosa; I, jointed chaeta, from a
Syllid; J, multidenticulate crotchet, from a Maldanid; K, comb-shaped chaeta,
from Eunice; L, uncinus of a Sabellid; M, uncinus of Terebellid (Amphitrite
Johnstoni); a, edgewise; b, side view; m, attachments of muscles into ba, basal
plate; x, accessory teeth. N, Sheathed spear of Hermione; a, the spear-shaped
capillary removed from its sheath; b, the same, with sheath.

Certain modifications of the chaetae presented by various worms deserve

mention. In Polydora (Fig. 133, A) and in Chaetopterus (Fig. 173, p. 324) those of
one segment are especially strong, but their significance is uncertain. In Capitella
those of the notopodium of the eighth and ninth segments are specially modified;
they are analogous to the copulatory chaetae of Oligochaeta. In Aphrodite, in
addition to the ordinary locomotor chaetae, there are brilliant, iridescent bristles
and peculiar felting threads arising from the indistinct notopodium; these latter,
however, are not true "chaetae," but are separate chitinous filaments similar to the
constituent fibres of an ordinary chaeta.[322]

While the chaetae in the Nereidiformia and others are grouped in bundles, those
of many other families are in vertical, transverse rows, as in Maldanidae and in
Arenicola. The uncini are always embedded in such rows, usually slightly raised
from the general level of the body surface, each being termed a "torus
uncinigerus." These tori are usually limited to the sides of the body, but in Myxicola
and in Notomastus they encroach upon the dorsal surface, and in Chaetozone,
also upon the ventral, so as nearly to encircle the body, recalling the
"perichaetous" condition of some earth-worms.

Fig. 139.—Aphrodite. Foot, × 2. a, Elytron; b, notopodium; c, neuropodium; d,

neuropodial cirrus; n, aciculum; 1, iridescent bristles; 2, stiff chaetae; 3, felt.

Gills.—We have already seen that several different organs, e.g. the palps in
Sabelliformia, the prostomial tentacles of Chlorhaemidae, and the notopodial cirri
of sundry other Polychaetes, may take on a respiratory function. There are,
however, certain "gills" developed either on the parapodium itself or elsewhere on
the body which it is difficult to homologise. Such are the retractile gills on the
parapodia of the Glyceridae (Fig. 136, C); those of Dasybranchus, near the
abdominal neuropodia; those of Mastobranchus, near the notopodia. Nephthys
has a sickle-shaped gill on the under surface of the notopodium. The long gill
filaments at the posterior end of Sternaspis, again, are only doubtfully interpreted
as the dorsal cirri of some of the posterior segments.

Since primitively the whole skin of the worm is respiratory, any part of the skin may
become more or less specialised for this function, and chiefly, of course, on the
more actively moving parapodia. The blood-vessels constituting the essential part
of the "gill" may make use of any already existing outgrowth (such as a cirrus or a
tentacle), or may push the body-wall out on their own account.

Internal Anatomy.

Probably those organs which have the greatest effect in modifying the shape of
the body are the septa, for we find in the long, free-swimming worms that these
are regularly present throughout the body, and external "segmentation" of the
body is well marked. In burrowing and tubicolous forms the septa are frequently
incompletely developed, or more or fewer may be absent; and the body becomes
less distinctly segmented externally, tends to vary greatly in diameter during
movement, or becomes plumper. With the disappearance of the septa there is also
a diminution in the number of nephridia, as in Arenicola, with only six pairs.
Further, there is frequently a dimorphism of these organs; instead of all of them
serving equally as excretory organs and as genital ducts, some of the most
anterior in the Sabelliformia and Terebelliformia become greatly enlarged, and
take on practically the whole of the former function; whilst more or fewer of the
posterior nephridia dwindle in size, and become genital ducts. The absence of
septa allows a free communication between the successive segments, and thus a
freer flow of coelomic fluid for the distension of the anterior end of the worm during
burrowing.

The alimentary system presents certain modifications of a systematic value. In

the Nereidiformia the muscular pharynx, which is always protrusible and is
preceded by an eversible buccal region, frequently encloses thickened cuticular
plates which serve as crushing and grasping organs. The form, number, and
arrangement of these "jaws" vary in the different families. They form valuable
fossil records of extinct worms.

In the Scoleciformia and Capitelliformia the buccal region exists, but there are no
jaws. In the Sabelliformia and Terebelliformia eversion does not take place and
jaws are absent.
Amongst the Nereidiformia the jaws are absent in the Phyllodocidae and
Hesionidae; when present they are usually set in the direct course of the food.
There may be one small tooth used for stabbing, as in some Syllids (Fig. 141, A);
or a circle of such denticles (Autolytus, Fig. 140, D). To these are added powerful
grasping jaws in Nereis (E); or the latter may alone be present, as in Glycera (F).
In Polynoë the four jaws are carried by hard pieces, to which the muscles are
attached (C and G). In Nephthys there is a dorsal and a ventral jaw.

Fig. 140.—Jaws of various Chaetopods. A, Transverse section of the anterior end of

Eunice; a, b, c, d, various parts of the upper series of denticles lying in a
special chamber; g, oesophagus; k, lower jaw: B, the denticles of Eunice
separated; U, upper series; a, grinder; b, forceps; c, rasping plates; d, grater; L,
lower series; j, tooth; k, base into which muscles are inserted: C, Polynoid; U,
upper, and L, lower jaws; j, tooth; k, base: D, Diagrammatic section across
pharynx of Autolytus; E, of Nereis; F, of Glycera; G, of Polynoë.

In the Eunicidae, however, the numerous denticles are carried in a special pouch
below the food tract, with which it communicates anteriorly.[323] They are arranged
in an upper and lower series. The lower series (L) consists of a pair of flat plates
(k) on each side partially embedded in and acted upon by muscles, with a harder
enamelled piece—the actual lower "tooth" (j)—at its anterior end. The upper series
(U) consists of several pieces, varying in shape and size in the various genera of
this family; but developmentally they result from modifications of two rows of small,
similar pieces.[324]

The intestine is generally straight and cylindrical, and is usually constricted by the
septa, if these are present. In the Polynoids the intervening sacculations become
so long as to receive the name of "caeca," which, in Aphrodite, become
enormously elongated (Fig. 142); there are eighteen pairs of them (c), each being
a slender tube bent upon itself, giving off short branches and dilated distally,
where it lies in the base of the parapodium.
 Fig. 141.—A, Alimentary canal of Syllid: B, transverse section of pharynx of the
same; b, buccal region; d, oesophageal outgrowth; g, salivary glands; i,
intestine; j, tooth; p, pharynx; s, gizzard: C, alimentary canal of Petta (after
Wirén); i, intestine; o, oesophagus; r, rectum; s, stomach.

Fig. 142.—Alimentary canal of Aphrodite. × 1. (From Gegenbaur.) a, Anus; b,

pharynx; c, caeca; o, mouth.

The intestine is looped in a few genera only, as in Trophonia, or coiled, as in

Sternaspis, Petta (Amphictenid, Fig. 141, C), and Ammotrypane. In the course of
the tube there may be a thick-walled muscular gizzard, with hard chitinous lining,
as in certain Terebellids, where it appears to replace, in function, the pharynx of
the Nereidiformia; in the Syllidae the gizzard is present in addition to the pharynx
(Fig. 141, A).

Glandular appendages of the oesophagus are present in many worms. Amongst

the Nereidiformia, the Syllidae and Hesionidae possess oesophageal diverticula
(Fig. 141, A, d), which are used, not for secreting a digestive fluid, but as
reservoirs for water and air swallowed by the worms; and are provided with
muscular walls, by which their contents can be driven out. They appear, in fact, to
be used like the swim-bladder of fishes.[325] Many Chaetopods take in water by
the anus—no doubt for respiratory purposes—and pass it forwards along the
intestine. In the Capitelliformia a special groove conducts the water for some
distance, then the groove becomes closed to form a canal, which, after a course
forwards as a free tube below the intestine, again enters the latter, constituting a
"siphonal apparatus," similar to that of the Echiuroids and the sea urchins.
Sense Organs.—In addition to the prostomial eyes, which are present in nearly all
the Nereidiformia and Spioniformia, eyes may exist elsewhere on the body: thus
Myxicola infundibulum and Fabricia possess a pair on the anal segment; in M.
aesthetica Clap. there is a pair to every segment; in Branchiomma there is a
compound eye near the tip of each gill filament (i.e. palp); whilst in Dasychone a
series occurs along each gill filament. All these examples belong to the
Cryptocephala, in which, owing to certain peculiar modes of life, these sense
organs are required in correspondingly peculiar positions. It is usually stated that
Polyophthalmus possesses, in addition to the usual prostomial eyes, twelve pairs
on as many successive segments; but the minute structure of these organs points
rather to their function as light-producing organs.

Fig. 143.—A gill filament, A, of Branchiomma, B, of Dasychone. a, Axis; f,

secondary filaments; o, compound eye; x, lappets.

The Capitelliformia and Opheliidae possess a pair of peculiar "ciliated pits" or

"nuchal organs" at the upper side of the head, between the prostomium and
peristomium, and capable of eversion (Fig. 144). They are most characteristically
developed in the Capitelliformia, where each organ abuts upon a special lobe of
the brain. The function of these "ciliated organs," which bear a great resemblance
to those of the Nemertines, is a matter of speculation. Similar organs, in the form
of simple pits or grooves, occur in many of the Nereidiformia, Terebelliformia, and
others.[326]

Otocysts are rare. Arenicola possesses a pair at the base of the prostomium, each
of which in some species retains an opening to the exterior.[327] They probably
serve as "organs of direction" rather than of "hearing." Aricia and Polyophthalmus
likewise have such organs on the prostomium; whilst Fabricia, Myxicola, Terebella,
and a few others possess them in the peristomium, or in some other segment of
the body.
 Fig. 144.—Ammotrypane aulogaster Rathke, enlarged. (From Cuningham.) Anterior
end. a, Prostomium; b, everted buccal region; c, notopodial cirrus; X, ciliated
organ everted; I, II, III, first three segments.

Reproductive Phenomena.—With a few exceptions mentioned below, the

Polychaeta are unisexual. The sexual cells are developed in all cases from the
lining epithelium of the body-cavity. The exact spot at which this occurs varies in
different cases; it may be, though rarely, on the floor of the body-cavity; it is more
usually on the wall of some blood-vessel, either the ventral vessel or on branches
of it; or on the many blind blood-vessels of Aphrodite. The number of such genital
organs is very great in most worms, but in those presenting two regions of the
body they are confined to the posterior segments (Sabelliformia, Terebelliformia,
Capitelliformia). The number is very limited in Arenicola and other worms
presenting but few nephridia: in the former genus there being six pairs, in
Trophonia only one pair.

The following genera are hermaphrodite:—Amphiglena, Salmacina, Protula,

Spirorbis, belonging to the Sabelliformia, to which must be added some
Hesionidae. In this family ova and spermatozoa are developed around the same
blood-vessel. But in the former group of worms (as also in Ophryotrocha) the two
kinds of cells are produced in different regions of the body. Thus in Protula the
anterior abdominal segments are male, the posterior ones female, while in
Spirorbis the reverse arrangement holds; and in Syllis corruscans the anterior
segments of the body contain eggs, whilst the posterior region contains
spermatozoa, and this region separates and becomes a male worm.

The eggs and spermatozoa in the Polychaeta are discharged into the sea either
by rupture of the body-wall or through the nephridia; the male and female
elements unite, and the resulting fertilised eggs undergo development, either
floating separately in the water, or embedded in jelly, or attached to the body or to
the tube of the worm.

The result of the segmentation of the egg is a free-swimming larva known as a

"Trochosphere," similar to that of Polygordius. The larvae of different species
present various more or less marked departures from this type, for instead of the
two girdles of cilia there may be only the anterior girdle, or there may be several
complete or incomplete girdles between the two typical ones, or there may be
(Chaetopterids) only a single girdle of cilia about the middle of the body, the two
typical girdles being absent.[328] The postoral region, after elongation, generally
becomes marked out into three segments, and these segments develop chaetae,
which are usually temporary and specially long.

The little animal is thus equipped for an independent life: the provisional chaetae
help in keeping it balanced; and in some cases (Spionidae) serve to protect the
little soft creature, for when it is touched it curls up, and its chaetae stick out at the
sides, so that it looks like a hairy caterpillar. But the larva is quite at the mercy of
the sea, for it is carried hither and thither by currents, and in this way the species
is disseminated. The larvae of the Polychaetes, like those of other animals, occur
at certain periods of the year in large quantities at the surface of the sea, and
serve as food for various larger animals.

Fig. 145.—A, Trochosphere of Nephthys. × 65. a, Anus; b, apical plate (brain); c,

apical tuft of cilia; c', girdle of cilia; i, intestine; m, mouth; st, stomach. B, Larva
of Spio, with three segments, eight days old. × 100. c, Preoral girdle of cilia; c',
preanal girdle; ch, long provisional chaetae; pr, prostomium with eyes. (From
Claparède and Metschnikoff.)

These larvae are at first very different from the adult animal, and the necessary
changes to be passed through are more or less great according to the species. It
is not our intention to describe these changes in detail.[329] The larva increases in
size, the permanent chaetae make their appearance in regular order, and the body
exhibits segmentation, the new segments always appearing just in front of the anal
segment. The internal organs gradually develop, and the prostomial and
parapodial appendages grow out in their turn. In the Sabelliformia the
multifilamentous "gills" arise by the continued branching of an at first simple
process (the palp) arising from the latero-ventral surface of each side of the
preoral lobe.[330] These gradually encroach dorsally and ventrally till the
prostomium is more or less encircled; meanwhile the peristomium grows forwards
so as to conceal the prostomium, which no longer increases at the same rate as
does the rest of the body.

Although most worms appear to discharge their ova directly into the sea and take
no further care of them, some make provision for their offspring either by laying
the eggs in a jelly, which will serve as food for the young larvae—Aricia, Ophelia,
Protula, Phyllodoce—or by attaching them to their body. In certain Polynoids the
eggs are attached by means of a secretion to the back, under the elytra, where
they undergo development up to a certain stage. In Exogone and some other
Syllids they are attached to the ventral cirri, or in Grubea limbata, all over the
back. In the female Autolytus (Sacconereis) a ventrally-placed brood sac is formed
by the hardening of a secretion; the eggs develop into embryos inside the brood
sac, and then become free, with head appendages and three pairs of parapodia.
Enormous numbers of such embryos may occur; for instance, some 300 were
counted in a brood sac of Autolytus ebiensis. In the case of tubicolous worms, the
eggs are frequently attached to the tube, either inside or outside. In Spirorbis and
Salmacina the operculum serves as a brood pouch.

Only a very few species are known to be viviparous, viz. Syllis vivipara Kr.,
Cirratulus chrysoderma Clap., Marphysa sanguinea Mont., and Nereis diversicolor
Müll.

In most genera there is no external difference between a mature worm filled with
generative products and an immature one, except, it may be, in the colour; for the
yolk of the eggs is frequently tinted yellow, or pink, or bluish, while the
spermatozoa in mass are white; so that the normal colouring of the worm may be
modified when filled with these elements. But in a few instances striking
anatomical peculiarities are exhibited by the mature worm.[331] In many species of
Nereis, for instance, those segments containing the generative products undergo
more or less extensive changes, while the anterior ones remain unaltered. The
body of the ripe Nereis is then distinguishable into an anterior non-sexual region
and a posterior sexual region; and so great are these changes in certain species
that the mature worms were for a long time believed to belong to a different genus,
and received the name Heteronereis. But we now know their true relations, thanks
to the work of Claparède and others. The males in the Heteronereid phase have
fewer unaltered anterior segments than the females, so that there is a sexual
dimorphism.

Fig. 146.—Male "Heteronereis" of N. pelagica L. × 1. A, Non-sexual region; B,

sexual, modified region. (From Ehlers.)

The changes which Nereis undergoes in its transformation affect chiefly (a) the
shape of the parapodia, and (b) the form of the chaetae of these parapodia. Other
organs may also be affected, though less noticeably; thus the eyes become
enlarged, the intestine may become so compressed by the generative products as
to be functionless, and the tail develops special sensory papillae.[332]

In the parapodia an increase in size and a sharper delineation of the various parts
take place; then flattened foliaceous outgrowths (Fig. 147, x, y) arise from certain
lobes of the feet, in which, too, the blood supply becomes greatly increased. The
old chaetae are pushed out by the development of new ones of quite a different
shape; these are jointed like the old ones, but the appendix is, in many species at
least, flattened and oar-shaped (Fig. 123, C, p. 246); and the chaetae are
arranged in a fan-like manner. Both these modifications are in evident relation to
the free-swimming habit which the Heteronereid now adopts. The new foot serves
as a swimming organ, the old one was a walking appendage.

Fig. 147.—Parapodium of male "Heteronereis" of N. pelagica L. × 10. (From Ehlers.)

a, Notopodial cirrus; b, notopodium; c, neuropodium with new chaetae; c',
foliaceous outgrowth; d, neuropodial cirrus; x, y, foliaceous outgrowths.

Whilst some species, such as the common British N. diversicolor, undergo no

change, and others become modified as just described, others, again, are
polymorphic. Claparède was the first to show that N. dumerilii may occur in at
least five different mature forms; these differ from one another in size, colour,
mode of life, character of the eggs, etc. The immature forms may become ripe and
lay eggs while still retaining the "Nereid" characteristics, or these immature forms
may become "Heteronereids"[333] whilst the sexual elements are ripening. There
are then three different kinds of males and of females in this one species, some
being found at the bottom of the sea, as the large Heteronereid form, while the
small Heteronereid swims on the surface. The relations of these various forms to
one another, and the causes leading to the assumption of a Heteronereid
condition in some cases and not in others, are unknown.

A somewhat similar phenomenon is exhibited by members of the family Syllidae.

[334] In this family sexual reproduction is frequently accompanied by the asexual
modes of fission and gemmation. In some genera, such as Eusyllis, Odontosyllis,
and Exogone, there occur changes quite similar to those characterising
"Heteronereis"—that is, the posterior segments in which the genital organs exist
become altered, so that the worm consists of two distinct regions, and is termed a
"Heterosyllis." The most marked change is the appearance of a dorsal bundle of
long capilliform chaetae in each of the genital segments (Fig. 148, I).

But in other genera the hinder genital region of the body becomes separated, on
maturity, from the anterior non-sexual region. Various stages of this "schizogamy,"
or fission into a sexual and a non-sexual zooid, have been observed in different
genera. In the genus Syllis the first segment of the sexual zooid, after its
separation from the asexual zooid, proceeds to bud forth a head. The character of
the head is alike in both sexes, though different species present heads of different
shapes; and as the worms were originally described as distinct genera, the names
then given are retained as descriptive terms. Thus the "Chaetosyllis" form has only
two tentacles; the "Ioda" form has three tentacles and a pair of palps. One and the
same species (e.g. S. hyalina) may successively pass through these stages.

With regard to the asexual portion, there is a regeneration of the tail segments
after the sexual zooid has separated; and the number of segments so regenerated
is usually equal to those that have become sexual. After a time these newly
formed segments will produce generative organs, and take on the characteristic
natatory chaetae, and this region will in its turn separate.

But in other genera, such as Autolytus, the regeneration of segments may

commence before the separation of the sexual zooid; and the head of the sexual
zooid becomes budded out before separation from the asexual portion. So that the
animal now consists of two worms, each with its own head, separated by a region
or zone of proliferation (Fig. 148, IV). Moreover, in some species not only is the
hinder part of the body converted into a sexual zooid, but the zone of proliferation
becomes very active, and produces by gemmation a large number of segments,
which become marked out, by the appearance of heads at intervals, into a number
of zooids, in which genital organs will later make their appearance. A chain of as
many as sixteen zooids may be formed in Autolytus (Fig. 148, V)—the hindermost
by conversion of the hinder part of the body of the original "stock," the intervening
zooids by gemmation.
 Fig. 148.—Diagrams illustrating the various stages in the asexual formation of a
chain of zooids. (Modified from Malaquin.)
I, Heteronereid or Heterosyllid stage. A, Non-sexual; A', sexual region of the body,
with modified parapodia.
II, Syllis. The hinder sexual region, B, is similarly modified, and will separate from
the parent zooid, A, and become an independent zooid.
III, Autolytus. The hinder zooid, B, develops a head by budding before separation.
IV, Autolytus, etc. A zone of budding (z) makes its appearance in front of the head of
B, and by its growth will give rise to a series of new segments in the middle of
the body.
V, Myrianida, Autolytus, etc. From this zone of budding a very large number of
segments have been formed, which have, further, become grouped so as to
form three individuals, C, D, E; B is the hindmost zooid, which is either formed
from the hinder segments of the parent zooid or is produced by budding, like C,
D, E.

One original "stock," or asexual zooid, thus produces several sexual zooids, but
these are only of one sex for a given stock. The males differ in several important
characters from the females; so different, indeed, are the two sexes that before
their history was worked out by Agassiz[335] they were placed in different genera.
The male zooid has thus come to be known as Polybostrichus (Fig. 149, B). It has
three tentacles and two bifid palps; there are two pairs of peristomial cirri; the
testes are confined to the four anterior segments, which are without natatory
chaetae. The female is termed Sacconereis, owing to the possession of a great
ventral brood sac; its head possesses no separate palps; the peristomium carries
only one cirrus on each side; ova occur in every segment of the body, and may
even extend into the hinder segments of the asexual zooid (Fig. 149, C).
 Fig. 149.—Myrianida fasciata. (From Malaquin.) The bright red markings of the living
animal are here represented black. A, An asexual individual which has
produced by budding from the zone (z) a chain of twenty-nine zooids, the
oldest being labelled 1, the youngest 29. B, A ripe male zooid (Polybostrichus),
with three tentacles and a pair of forked palps (p). There are five unaltered
anterior segments. C, A ripe female zooid (Sacconereis) with the palps fused
with the prostomium; s, the ventral brood pouch projecting on each side; t,
tentacles.

A further development of this process of gemmiparity is exhibited by Myrianida.

Here, there is no conversion of the hinder segments, but the normal preanal zone
of proliferation gives rise to a large number of new segments. After a time the
most anterior of these becomes a head, and thus a new zooid is marked out. The
zone of proliferation immediately in front of the new head now proceeds to form
new segments, and a second zooid results. This process goes on till a
considerable number of new worms have been formed at the tail of the original
one, the oldest of these new ones being the most posterior, the youngest next the
original "stock." In each zooid there is a zone of activity which adds to its number
of segments, so that as we pass backwards the zooids increase in size. As many
as twenty-nine such zooids may be formed in this way entirely by gemmation; and
as each zooid becomes completed, genital organs make their appearance, and
when these are ripe the zooid separates from the "colony" and leads an
independent life. Here, as in Autolytus, the sexes are dimorphic, the male and
female resembling those of that genus.

The process of gemmation, as seen in Autolytus, closely resembles that exhibited

by certain Oligochaeta (Naididae), where there exists a definite alternation of
generations; the production of new individuals by gemmation occurring throughout
the greater part of the year, and sexual reproduction recurring only at certain
intervals. In the Polychaeta such alternation exists in Myrianida; but it is only the
terminal link of a series, which takes its starting-point in the process exhibited by
the majority of Annelids, where no sexual character marks maturity. The next
stage is presented by "epigamous" forms like Heteronereis and Heterosyllis; then