Skip to main content

Full text of "Alaska"

See other formats


Bv mmm 

i llii' 



n? Wrrn !; U'.li 

iirnin-:;H;n!:;v|;iiiiUh"'i;i!i;i'iiii': 1^1 





Digitized by tine Internet Arciiive 

in 2007 witii funding from 

IVIicrosoft Corporation 







John Andrew & So 

Bryn Mawr. Glacier 



I ( 




(Publication 1992) 






The publication of the series of volumes on the 
Harriman Alaska Expedition of 1899, heretofore pri- 
vately printed, has been transferred to the Smithsonian 
Institution by Mrs. Edward H. Harriman, and the 
work will hereafter be known as the Harriman Alaska 
Series of the Smithsonian Institution. 

The remainder of the edition of Volumes I to V, 
and VIII to XIII, as also Volumes VI and VII in 
preparation, together with any additional volumes that 
may hereafter appear, will bear special Smithsonian 
title pages. 

Smithsonian Institution, 

Washington, D. C, July, 1910 








Copyright, 1904, 


Edward H. Harriman. 


The present account of Alaska Glaciers, by G. K. Gil- 
bert, forms the third of the Harriman Alaska Expedition 
volumes and the first of the technical series. It is based 
on observations made by the author in the summer of 
1899, while a member of the Expedition, supplemented 
by information and photographs from various sources, and 
is illustrated by 106 text figures and 18 plates. The vol- 
umes on Geology, Cryptogamic Botany, Insects and Crus- 
taceans are to appear simultaneously with it. The vol- 
umes on Nemerteans, Oligochetes, and several other 
groups of Invertebrates, are ready for the press, and will 
follow as soon as they can be printed. The two volumes 
on Phanerogams, or flowering plants, are expected to ap- 
pear early in the year 1904. 

Acknowledgments are due the Canadian International 
Boundary Commission, the United States Biological Sur- 
vey, and, especially, the United States Geological Survey, 
for the use of photographs and other illustrative material. 
The editor wishes also to express his appreciation of the 
drawings prepared for this volume by Mr. W. E. Spader 
and Mrs. Louise M. Keeler. The frontispiece, plate xi, 
and a large number of the text figures are by Mr. Spader, 
while all of the chapter headings are by Mrs. Keeler. 

C. Hart Merriam, 

Washington, D. C, 
May /, igo3. 




Introduction i 

Route 3 

Photographs 4 

Chapter I. — Existing Glaciers 7 

General Distribution 7 

Lynn Canal 11 

Glacier Bay 16 

Muir Glacier 20 

Reid Inlet 25 

Bergs 33 

Hugh Miller Inlet 34 

Geikie Inlet 38 

La Perouse Glacier 39 

Yakutat Bay 45 

Hidden Glacier 52 

Nunatak Glacier 58 

Hubbard Glacier 63 

Turner Glacier 66 

Osier Island 69 

Columbia Glacier 71 

College Fiord 81 

Harriman Fiord 89 

Grewingk Glacier 97 

Summary of modern changes 102 

Theories 1 06 

Chapter II. — Pleistocene Glaciation 113 

Hanging Valleys .... 114 

District of Inland Passages 119 

Pre-Pleistocene topography 122 

The High Peneplain 123 

Low Peneplains 130 

A lower Base-level 134 

^ Summary 139 



Chapter II. — Pleistocene Glaciation (continued). page. 

District of Inland Passages (continued) . 

Glaciation 139 

Rounding of angles 139 

Cirques 141 

Fiords and Hanging Valleys 142 

Inequality of glacial erosion 159 

Glacial deposits 161 

Associated sea-levels 162 

High Mountain district 166 

Prince William Sound 173 

Kenai Pe nins ula 177 

Kadiak Island 177 

Region of the Gulf Coast 182 

Unalaska Island 185 

Bering Sea ^ 186 

Chapter III. — General Considerations as to Glaciers 195 

The surface of a glacier 196 

Evenness as compared to rock floor 196 

Lateral cliff 197 

Crevasse cycle 198 

Glacial sculpture 203 

Pressure and erosive power of Tidal Glaciers 210 

Rivers of ice and of water 218 

Resemblances 218 

Differences 220 

Homologies and Analogies 220 




I. Bryn Mawr Glacier Frontispiece 

II. Map of Alaska, showing route i 

III. Hugh Miller Glacier 36 

IV. Map of Hidden Glacier 52 

V. Kettle-holes 54 

VI. Profile of Hidden Glacier. Section of moraine 56 

VII. Map of Nunatak Glacier 58 

VIII. Map of Hubbard and Turner glaciers 62 

IX. Panoramas of Hubbard and Columbia glaciers 64 

X. Turner Glacier in 1 89 1 and in 1899 66 

XI. Map of Columbia Glacier 70 

XII. Columbia Glacier, from Heather Island 74 

XIII. Maps of Prince William Sound and Port Wells 80 

XrV. Amherst and Crescent glaciers.. 82 

XV. Harriman Glacier 94 

XVI. A hanging valley, Fraser Reach.. 116 

XVII. Lowlands near Wrangell and Sitka 132 

XVIII. Glaciated rocks 206 



1. Distribution of glaciers in Alaska 8 

2. Davidson Glacier, from peninsula.... 12 

3. Davidson Glacier, from above , 13 

4. Davidson Glacier, side view 14 

5. Map of Davidson Glacier 15 

6. Longitudinal section of Davidson Glacier 15 

7. Map of Glacier Bay 18 

8 . Remnant glacier south of Hugh Miller Inlet 20 

9. Muir Glacier 20 

10. Map of Muir Inlet 21 

11. Map of Reid Inlet. 27 




12. Reid Glacier, from the north 28 

13. Reid Glacier, from the northeast 29 

14. Reid Glacier, from the northwest 30 

15. Reid Glacier, from the northeast' 31 

16. Rock-laden berg 33 

17. Map of Hugh Miller Inlet 34 

18. Till left by Hugh Miller Glacier 35 

19. Charpentier Glacier , 36 

20. La Perouse Glacier 39 

21. Map of margin of La Perouse Glacier 40 

22. Section of timbered ridge near La Perouse Glacier 41 

23. Barren zone at margin of La Perouse Glacier 41 

24. Push-moraine of La Perouse Glacier 42 

25. La Perouse Glacier ; contact with forest 43 

26. Map of Yakutat Bay 46 

27. Map of Yakutat Bay, showing configuration of bottom 50 

28. Hidden Glacier 53 

29. Waste plain of Hidden Glacier , 57 

30. Westward from Nunatak Glacier 59 

31. Tidal front of Nunatak Glacier 60 

32. Cascading glacier in Nunatak Fiord 61 

33. South tongue of Nunatak Glacier 61 

34. Tidal front of Nunatak Glacier 62 

35. Hubbard Glacier 64 

36. Longitudinal section of Turner Glacier 68 

37. Panorama of Columbia Glacier 72 

38. Western edge of Columbia Glacier 76 

39. Push-moraine, west shore of Columbia Bay 77 

40. Fluted moraine at edge of Columbia Glacier 77 

41. Moraine marking attack by Columbia Glacier on forest 79 

42. Outlines of Columbia Bay 81 

43. East part of front of Yale Glacier 83 

44. Harvard Glacier 8^ 

45. West side of College Fiord 87 

46. Distant view of Barry Glacier 90 

47. Part of front of Barry Glacier 91 

48. Caves in frontal cliff of Barry Glacier 91 

49. East part of front of Barry Glacier 92 

50. Serpentine Glacier 93 

51. Surprise and Cataract glaciers 94 



52. Hemlocks bordering Harriman Fiord.. 97 

53. Grewingk Glacier 99 

54. South wall of valley at front of Grewingk Glacier 100 

55. Distribution of glacier localities 105 

56. Diagram illustrating origin of hanging valleys 115 

5 7 . Diagram illustrating origin of hanging valleys 115 

58. Diagram illustrating discordance of hanging valleys 116 

59. Map; fiords of the Inland Passage district 120 

60. An Alaska fiord 121 

61 . Upland topography near Cape Spencer 1 25 

62. Upland topography near Berner Bay 127 

63. Upland topography near Walker Bay , 1 28 

64. Part of Annette Island 131 

65. Old peneplain near Sitka 132 

66. Douglas Island, from Juneau 133 

67. A fiord of the Inside Passage 141 

68. A fiord of the Inside Passage 142 

69. Hanging valley on Vancouver Island 143 

70. Hanging valley on Princess Royal Island 145 

71. Hanging valley, Fraser Reach 146 

72. Hanging valley, Fraser Reach 146 

73. North wall of Lowe Inlet 148 

74. Head of Lituya Bay 149 

75. Head of Lynn Canal 152 

76. Hanging glacier, Taiya Inlet 153 

77. Northeast wall of Chilkoot trough 154 

78. Profile from Taiya Pass southward 155 

79. Cross-profiles about head of Lynn Canal 155 

80. Mouth of Speel River 158 

81. Hanging valley, Nunatak Fiord 171 

82. Peninsula near Orca 174 

83. Hinchinbrook Island, from the sea 174 

84. Diagram of northwest wall of College Fiord 175 

85. Terrace on Spruce Island, opposite Kadiak Island 179 

86. Islands off Kadiak 180 

87. Hill behind Kadiak village 181 

88. Hanging valley near Kadiak village 181 

89. Crater rim on St. Paul Island 187 

90. Unglaciated knob on St. Matthew Island 189 

91. Stream-graded valley near Port Clarence 189 




92. East wall of Plover Bay, Siberia 190 

93. Hanging valley, Plover Bay 191 

94. Downward limit of crevasses in Muir Glacier 1 98 

95. Crevasses and seracs, Muir Glacier 199 

96. Pinnacles on Columbia Glacier 2cx> 

97. Back of Columbia Glacier 200 

98. Level tract on Muir Glacier 201 

99. Ideal profile and section of a glacier 201 

100. Ideal section of water-filled crevasses 202 

loi. Influence of drift on surface character of glacier 202 

102. Ice sculpture in Russell Fiord 209 

103. Diagrams illustrating flotation theory of tidal glaciers 210 

104. Diagrams illustrating non-flotation theory of tidal glaciers... 211 

105. Contact phenomena of glass and mercury 212 

106. Ideal section of tidal glacier 215 




The glaciers of Alaska are many and the district 
through which they are distributed is large. The region 
formerly overrun by them is still larger. Ten years 
ago Russell and Kerr surveyed and studied Malaspina 
Glacier and its dependencies, and about the same time 
Reid made a scientific survey of Glacier Bay and its 
bordering glaciers; but these two tracts are only dots on 
the general map. All other glacial studies in the great 
field have been of the nature of reconnaissance and most 
of them have been carried on incidentally in connection 
with general geographic or geologic work. The Harriman 
Expedition added one more to the list of reconnaissances. 

Until recently observations have been limited to the 
coast and it is still true that the greater number of glaciers 
which have been studied or mapped are coastal, or at least 
visible from the sea; but at the present time interior gla- 
ciers are receiving more attention, and knowledge of them 
is rapidly growing. Geographic investigation, so long as 



stimulated by only the fur and salmon industries, was di- 
rected chiefly to coastal regions, but the development of 
gold placers turned attention to the interior, and for sev- 
eral years the topographers and geologists of the United 
States Geological Survey have been actively engaged in 
the exploration and survey of inland districts. 

The growth of knowledge of Alaska glaciers is so rapid 
that a summary of existing knowledge would have but 
transient value. Generalization as to most points of world- 
wide interest is at present impracticable because observers 
have differed widely in their preconceptions and criteria, 
and the data consequently are not homogeneous. It has 
seemed best, therefore, to make the present report pri- 
marily a record of the data gathered by the Harriman Ex- 
pedition and to make use of other material only when it is 
closely connected with the new data or is otherwise ser- 
viceable in their interpretation. 

Regarded as a reconnaissance of glacial geology, the 
cruise of the * Elder ' was fairly comprehensive. It not only 
covered rapidly a wide extent of coast but it brought 
under view a great variety of phenomena. The general 
impressions acquired while the ship was skirting the coast 
were supplemented by the results of more definite and 
detailed observation at a few points on the land; and the 
impressions, acquired by local studies of individual glaciers 
were enlarged by the panoramic view of many others. 
Opportunities for close examination included landings 
from the ship at thirty-four localities, at three of which 
the use of a camping outfit extended the time to several 
days. The remainder of the two months covered by the 
voyage was spent on the ship, and about half the sailing 
time was so conditioned by distance from shore, by light, 
and by weather, as to permit profitable observation of the 
coast. After the voyage was over physiographic studies 
were continued by the aid of photographs. Thousands of 


views by members of the Expedition were examined, as 
well as a large number from other sources, and at least 
several hundred of these have yielded information as to 
glaciers and glaciation. 

In arranging the material for presentation it has been 
found convenient to make an arbitrary division of the 
history of glaciation, connecting such changes as appear 
to have occurred within a few hundred years with the ex- 
isting status, and classing all remoter changes with the 
geologic or Pleistocene series. This procedure is a matter 
of convenience only; it is not determined by a turning 
point in glacial history, but by a difference in the nature 
of the evidence by which the history is recorded. The 
direct observational record, for a few localities, reaches 
back a little more than a century, and inference from the 
age of trees extends a little farther; but for all earlier 
times the data are purely geologic and the changes have 
not been measured in years. 

Under this classification the heads of my principal chap- 
ters are Existing Glaciers and Pleistocene Glaciation, 
The changing relations of sea and land also receive atten- 
tion, but these are so closely connected with the problems 
of Pleistocene glaciation that they have not been given a 
separate place. Notes of a general character as to glaciers 
and their work are in part introduced along with local 
descriptions and in part assembled in a closing chapter. 

Route, — Through the greater part of the journey I re- 
mained with the main party, so that the red line on the 
route map (pi. i) shows my course with approximate 
accuracy. It seems necessary to mention here only a few 
deviations and details. Such dates as are of importance 
are noted in connection with the descriptions of individual 

In Glacier Bay I spent a day and a half at Muir Gla- 
cier, and then, with Muir and Palache, visited Hugh Miller, 


Reid and Geikie inlets in a rowboat. This excursion oc- 
cupied four days, three camps being made on the shores 
of the bay. 

In Yakutat Bay I landed at Hidden and Nunatak gla- 
ciers and at the summer village of the Yakutat Indians. 
Two days were spent, with Muir, Gannett, and Kearney, 
in a boat excursion which visited Hubbard Glacier and 
Osier and Haenke islands. 

In Prince William Sound I touched at Orca, and was 
then left at Columbia Glacier with a boat and shore party, 
including Palache, Coville, and Curtis, while the ship ex- 
plored Port Wells. We remained three days, and after- 
ward had a partial view of Harriman Fiord. 

In Cook Inlet I was landed for a half day at Grewingk 
Glacier, while the ship made an excursion up the main 

I was with the ship on all routes about Kadiak Island, 
landed briefly on the western coast, spent several days at 
and near Kadiak village, and visited Long Island. 

At Port Clarence I was of a party that crossed the bay 
in a launch and visited the mainland. 

Photographs, — For the study of changes in the size 
of glaciers photographic views are of peculiar value. A 
view showing a glacier or part of a glacier in relation to 
details of adjacent land constitutes a record which can at 
any time be compared either with the objects themselves 
or with another photograph made in another year or 
month. That a photograph may have its highest value 
for such use its date must be known, including year, 
month, and day of month. The Harriman Expedition 
carried many cameras and secured a large number of 
views of glaciers. Some of these views are reproduced 
in the present and preceding volumes, and are thus made 
available for the investigator. So far as they are contained 
in this volume, their dates are given in the associated text. 


The geographer who shall undertake in the future to study 
the variation of Alaska glaciers will have use not only for 
the fuller details of the original photographs but for a 
more complete series, and the following information is 
given for his benefit. 

Of the many series of photographs made by the mem- 
bers of the Expedition, three may be regarded as public, 
and these also contain the most important records of the 
glaciers. Mr. E. S. Curtis, the official photographer of 
the Expedition, made the largest and best views, and keeps 
them on sale at his photographic establishment in Seattle, 
Washington. The negatives made by Dr. C. Hart Mer- 
riam are filed at the office of the Biological Survey, U. S. 
Department of Agriculture, Washington, D. C. My own 
negatives are filed at the office of the U. S. Geological 
Survey, Washington, D. C, where a set of prints may be 
seen and prints from negatives may be ordered. 

The dates of our photographs for the several glaciers 
are as follows, the year 1899 being understood in each 
case: Amherst, June 26; Barry, June 26-29; Bryn Mawr, 
June 26; Cataract, June 27; Charpentier,June 11; Colum- 
bia, June 25-28; Crescent, June 26; Crillon, July 24; 
Davidson, June 6; Grand Pacific, June 12; Grewingk, 
July 21; Harriman, June 27; Harvard, June 26; Hidden, 
June 20; Hubbard, June 19-22; Hugh Miller, June 11; 
Johns Hopkins, June 12; La Perouse (near views), June 
18; La Perouse (distant views), June 18 and July 24; 
Muir, June 8-13; Nunatak, June 20; Radcliflfe, June 26; 
Reid,June 12; Roaring, June 27; Serpentine, June 26-27; 
Smith, June 26; Surprise, June 27; Turner, June 19-22; 
Vassar, June 26; Wellesley, June 26; Yale, June 26. 

The table on page 6, giving series and negative num- 
bers of photographs from which text figures were made, 
includes most of the figures based on photographic views. 
Information as to others is given in their labels and other 


associated text. Titles of series are abbreviated in the 
table as follows : ESC = E. S. Curtis, Seattle, Wash.; 
USGS = U. S. Geological Survey, Gilbert; USES = U. 
S. Biological Survey, Merriam, 1899; CBC = Canadian 
Boundary Commission, Ottawa, Canada. The Boundary 
Commission photographs are mounted in an album of 
many volumes, and the complete reference includes num- 
ber, volume and page. A copy of this album is in the 
library of the State Department at Washington. 




Negative or Photo. 



Negative or Photo. 



3, V. 12, p. 8 






117, V. 12, p. 43 



93-4, V. 14, p. 53 






96, V. II, p. 30 




132-3, V. 8A, p. 50 









44, V. 14, p. 13 
38, V. 14, p. II 


















































109A, V. 12, p. 81 






93-4, V. 12, p. 3 











68, V. 12, p. 42 






93, V. 6, p. 43 
























398, 400 























393 (?) 




































3» V. 17, p. 3. 







Nearly all the glaciers of Alaska are comprised within 
a belt of moderate width which follows the southern coast 
from the Aleutian Islands to Portland Canal (see fig. i). 
Curving about the great bight of the Pacific Ocean known 
as the Gulf of Alaska, this belt has a length of i,6oo miles, 
and its extreme width, near the middle, is about 250 miles. 
Within it the arrangement of glaciers is irregular, but their 
more important groups occupy the middle region, while 
near the ends they are comparatively sparse and small. 

The explanation of this massing of glaciers along the 
southern coast is not far to seek. The general circulation 
of the Pacific Ocean brings to the Gulf of Alaska a cur- 
rent of water which has been warmed in the tropics and 
still retains so much heat that its mean temperature is 
considerably above the normal for the latitude. The 
ocean is therefore, at most seasons, warmer than the con- 
tiguous land, and though air currents passing from ocean 
to land convey heat to the land they are themselves cooled. 




While traversing the ocean the air becomes loaded with 
moisture, the cooling over the land diminishes its water- 
carrying capacity, and part of its load falls to the ground as 
rain or snow. Moreover, all this coast is mountainous, so 
that landward flowing air is compelled to rise, and its capac- 
ity is still further reduced by rarefaction. At the greater 
altitudes the ratio of snow to rain is comparatively large, 

The belt containing glaciers is shaded by parallel lines. 

and the mountains thus become gathering grounds for the 
snows that feed glaciers. Farther inland the air currents 
descend somewhat and the precipitation is diminished until 
the conditions for glacier formation cease. Hayes states 
that while the neve line of glaciers on the southward face 
of the St. Elias Alps lies at about 2,000 feet above sea- 
level, its altitude on the northern face is over 6,000 feet.^ 

» An Expedition through the Yukon district. By Charles Willard Hayes. 
Nat. Geog. Mag., vol. iv, p. 153, 1892. 


Along the western coast of Alaska the conditions are 
different. Bering Sea lies practically outside the influence 
of the Pacific circulation and the temperature of its water 
is approximately normal. Its power to charge air cur- 
rents with moisture is small, especially in winter; and 
though the winter temperature over the adjacent land is 
low, the snowfall is not heavy. There are no great moun- 
tain ranges to concentrate the precipitation, and the snow 
of winter, being broadly spread over plains or caught by 
ranges of moderate height, is dissipated by the melting 
and evaporation of summer. 

The glacier-bearing belt includes about three-tenths 
of the vast territory of Alaska. Its exploration has but 
begun, yet enough is known to give it rank as the third 
great glacier district of the world, only the Antarctic con- 
tinent and Greenland surpassing it. Its ice may be 
roughly estimated to occupy a tenth of the surface, or an 
absolute area of between 15,000 and 20,000 square miles, 
and this expanse is so divided and scattered as to offer to 
the student the utmost variety of local condition and de- 
tail. Of alpine glaciers, such as would receive individual 
names if near the homes of men, there are many hundreds, 
possibly more than a thousand; of broad, composite fields, 
like the Muir and Malaspina, there are about a half dozen; 
and more than thirty are known to reach the coast and 
cast bergs into the sea. 

Skirting the land in a ship and making only brief ex- 
cursions away from it, we saw only glaciers of the coastal 
mountains and lowlands, and close inspection was limited 
to the lower ends of ice streams. The phenomena which 
arrested our attention were those of wasting, of the depo- 
sition of detritus, and of the advance and retreat of the ice 
fronts. It soon became evident that our chief opportunity 
to advance glacial science was through contributions to 
the history of local changes in the frontal boundaries of 


glaciers. That which was accessible to us had been 
accessible to our predecessors also, so that at several 
points we could compare present with past condition ; 
and, for like reason, whatever record we might make 
could readily be used by the investigators of the future. 
Effort was accordingly made to visit as many as possible 
of the glaciers already described and mapped, and at all 
points visited to secure an intelligible record of the exist- 
ing status. 

The plural pronoun in the preceding paragraph is 
not the conventional affectation of modesty, but springs 
naturally from the consciousness that the facts I am 
to present were not wholly of my own observation. In 
grouping the material for publication it seemed to my 
colleagues in geology, Emerson and Palache, as well 
as to myself, that it would be better to classify by sub- 
jects than by observers, and as glaciers fell to my 
share, I have absorbed the glacial observations made 
by these gentlemen. I am greatly indebted also to the 
map work of Gannett, to the historical data and fertile 
suggestions of Muir, and to the timely cooperation of 
Dall and Coville. 

Before taking up the description of the glaciers, a few 
words will be devoted to the terminology connected with 
their broader classification. The distinction between aU 
'pine glaciers (sometimes called glaciers proper) and con- 
tinental glaciers (also called ice-sheets) has long been 
recognized. Alpine glaciers are fed by neves in high 
mountains and as rivers of ice descend mountain valleys. 
Continental glaciers gather on broad plains or plateaus 
and spread outward. Russell, as a result of studies in 
Alaska, recognized a third type, the piedmont} A pied- 
mont glacier is a broad sheet of ice resting on a lowland 

»An Expedition to Mount St. Elias, Alaska. By Israel C. Russell. Nat. 
Geog. Mag., vol. m, p. 121, 1891. 


at the foot of a mountain range, resembling a continental 
glacier in its mode of wasting, but distinguished by the 
fact that it is fed by the alpine glaciers of the mountain. 
An alpine glacier may be simple and separate, or com- 
pound. Two or more often descend from the same neve. 
Still more frequently two meet and coalesce after the 
manner of rivers, and a trunk glacier may have many 
tributaries. Small alpine glaciers are sometimes called 
glacierets^ or, if visible high on the sides of mountain 
valleys, hanging glaciers. 

For the purpose of the present report it is convenient to 
distinguish glaciers which reach the sea and discharge 
bergs, from those which end on the land. Reid calls 
these, severally, tide-water and alpine glaciers;^ and 
they have also been called (after whose initiative I do not 
know) live and dead glaciers. Reid's use of * alpine ' con- 
flicts with the well-established use already mentioned, and 
the terms ' live ' and ^ dead ' are clearly misleading, as the 
great majority of the active glaciers of the world fail to 
reach the sea. I shall abbreviate ^ tide-water ' to tidal and 
employ non-tidal as its antithesis. 


The order in which the glaciers were observed was from 
east to west, and it has been found convenient to adopt 
this as the order of description also. 

On the islands of southeastern Alaska there are no 
glaciers, and those of the mainland, south of Juneau, nestle 
in recesses of the mountains so far from the steamer route 
that we had only distant glimpses. But in Lynn Canal we 
followed a great fiord between ranges at once so lofty as 
to project well above the snow-line and so bold as to ex- 

1 Glacier Bay and its Glaciers. By Harry Fielding Reid. Sixteenth Ann. Rept. 
U. S. Geol. Survey, part i, 1896. See pp. 429 and 442, The term tide-water 
was used by Russell as early as 1892. See Am. Geol., vol. ix, p. 322, 1892. 



hibit their crowning banks of snow and ice in continuous 
panorama. These ranges have been mapped in some de- 
tail by the U. S. Coast Survey and the Canadian Inter- 
national Boundary Commission, and it appears from a 
comparison of the glaciers with contours of altitude that 
the snow-line descends in both directions away from the 
fiord. On the mountain slopes overlooking the water 
glaciers do not form at a lower height than 4,500 feet, but 


Showing the trench form of its valley ,'the spreading of its end, and the two zones of the 
fringing plain. Photograph by W. Ogilvie, from peninsula at left in fig. 3. See page 6. 

on the opposite slopes of the same ranges 3,500 feet seems 
to be enough. 

Davidson Glacier, fed by a high snow-field lying several 
miles back of the first mountain crest on the west, flows 
to the fiord through a narrow trench and reaches sea- 
level, though its ice does not actually touch the water of 
the ocean. In its mountain trench it has a width of only 
a half mile, but on escaping from the confining rock walls 
and entering the fiord it immediately spreads into a semi- 
circular fan with a radius (in 1 894) of about three-fourths 
of a mile. All about its frontal margin is a fringe of low- 
land averaging three-fourths of a mile in breadth, consti- 
tuted, at the surface, of rock waste brought by the glacier. 



There may be much unmelted ice beneath the visible 
gravel, but the lowland certainly contains a great body of 
gravel which, if glacier and ocean were withdrawn, would 
appear as a great curved ridge of water-laid moraine stuff. 
Glacier and moraine together encroach nearly two miles 
on the water of the fiord — a branch of Lynn Canal called 
Chilkat Inlet — reducing its width to about one mile. 


View of terminal fan from the high peak at the right in fig. 2. Shows the barren and 
forested zones of the fringing plain. The glacier is partly concealed by a rock of the fore- 
ground. From a photograph by J. J. McArthur, 1894, See page 6. 

The depth of the moraine, or moraine-delta, where it 
occupies the middle of the fiord is more than 500 feet. 
The profile of the ice fan, as shown by photographs, has 
a slope of one foot in ten (see fig. 4). The gravel low- 
land is much flatter, but the submerged face of the deposit 
descends to the bottom of the fiord with a general rate of 
one foot in three. 

We made no landing here, and our facts are derived 
chiefly from Russell, who visited the foot of the glacier in 
1889; from the Coast Survey chart, based on surveys in 
1890-94; and from photographs made by the Canadian 



Boundary Commission in 1894.^ They are of interest in 
this connection chiefly from their bearing on the interpre- 
tation of morainic ridges observed farther west, at the front 

Side view of expanded end, June, 1899. 

of the Fairweather Range. If Davidson Glacier were to 
melt away and the sea to retreat from Chilkat Inlet, there 
would remain in the valley a high ridge of water-laid 
gravel, recording by its circling course the present out- 
line of the glacier front, and by its even crest line the 
present position of the sea surface. 

As to modern changes in the extent of this glacier there 
is little specific information, but the condition of the fring- 
ing plain warrants a few general statements. An outer 
zone is covered by forest, an inner is barren (figs. 2, 3 and 
5). The forest is in general lofty, dense, and apparently 
mature, but a narrow belt next the barren zone has smaller 
trees in rather open growth. The forest zone has an 
average width of less than half a mile, ranging from one- 
fourth at the north to five-eighths at the south; the bar- 

^See page 6 and figs. 2 and 3. 



ren zone in 1894 ranged from one-fourth mile (east side) 
to nearly one-half mile (north side). There were many 
lakelets in the barren zone (1894) and a few in the forest, 


Based on Coast Survey chart 8303 and Boundary Commission photographs made in 1894. 
The contour interval on land is 500 feet ; under water, 250 feet. Barren lowlands dotted ; for- 
est indicated by stars. 

and these are probably indicative of the melting of ice 
buried under the gravel. It seems safe to infer (i) that 
for centuries (age of the mature forest) the glacier has 
not exceeded its present extent by more than three- 
eighths of a mile, and (2) that the period since it last 



g, glacier, m, moraine-delta, w, water of inlet, p, profile of mountain range near glacier. 


reached the three-eighth-mile limit is roughly measured 
by the age of a half-grown forest. 


The next depression west of Lynn Canal is broader, 
as well as more complex in the details of its config- 
uration, and its general trend is northwestward in- 
stead of northward. On the west it is separated from 
the Gulf of Alaska by the Fairweather Range, the 
dominant mountain mass of this region. Glacier Bay, 
occupying its main axis, sends many branches into 
the troughs among its hills and mountains, and would 
be still more complicated in outline but for the clog- 
ging of the valleys by glaciers. Six of its inlets head 
against ice walls, from which bergs are constantly 

The glaciers of this basin are better known to the phys- 
ical geographer than any others of the Alaska belt, and 
one of their number, the Muir, enjoys the same preemi- 
nence in popular acquaintance and appreciation. The 
group was explored, sketched, and studied by Muir in 
1879 and 1880. Wright made a visit in 1886, producing 
a rough map and working out the main elements of the 
later glacial history. Reid, in 1890 and 1892, executed 
a careful survey of the shores, the lower portions of the 
tidal glaciers, and a broad inland tract including the 
greater part of Muir Glacier; developed much further 
the history of recent changes; and instituted a number 
of investigations bearing on the physics of glaciers. 
Minor studies were made by Gushing (1890) and Rus- 
sell (1890), the map work has been extended by the 
Canadian Boundary Commission, and a fine series of 
photographs were made by the Commission in the sum- 
mer of 1894. 


To the body of information gathered and published by 
these investigators ^ our addition was comparatively unim- 
portant. The five long days of our stay, though utilized 
to the utmost and replete with interest, served only for a 
partial review of the features of the immediate coast. As 
to the more general aspects of the topography and physical 
history we verified the work of our predecessors, but we 
were able to extend it only by bringing down to date the 
records of changing glacier fronts. 

Except near the mouth, the shores of Glacier Bay are 
treeless, and large tracts are almost destitute of vegetation. 
A variety of other facts show that this barrenness is due 
to the recent occupation of the surface by ice, and the ex- 
tent of the barren zone measures the amount of modern 
recession of the glaciers. Another series of phenomena, 
including vestiges of a forest and remnants of a moraine- 
delta, show that the epoch of expanded glaciers was pre- 
ceded by an epoch of contracted glaciers when the ice 
occupied less space than at present. Thus, from a condi- 
tion of minimum extent, the glaciers grew to a maximum 
of brief duration and then wasted away to their present 
dimensions. Vancouver's narrative seems to show that at 
the time of his visit to the coast (1794) the ice was near 
its maximum, and subsequent observations, beginning with 
those of Muir in 1879, show rapid and nearly continuous 
retreat. The magnitude of this oscillation is perhaps 
without parallel in the records of glacial changes within 
the historic period. During the maximum epoch an ice 
flood not only filled Glacier Bay for thirty-five or forty 
miles, but submerged many islands and bordering hills 

ijohn Muir, Cruise of the Corwin in the Arctic Ocean, p. 136, 1884; The 
Mountains of California, pp. 23-24, 1894. H. F. Reid, Nat. Geog. Mag., vol. 
IV, 1892 ; Sixteenth Ann. Rept. U. S. Geol. Survey, part i, 1896. H. P. Cushing, 
Am. Geol., October, 1891. G. F. Wright, Am. Jour. Sci., Jan., 1887; The Ice 
Age in North America; Man and the Glacial Period. I. C. Russell, Am. Geol.» 
March, 1892. 



from 1,000 to 2,000 feet in height and reduced the lesser 
mountains of the basin to the condition of projecting peaks, 
or nunataks. The sea front made a continuous ice cliff 
ten miles long. Then the great trunk glacier gradually 

wasted; the rounded 

crests of submerged 
hills began to reap- 
pear as nunataks ; 
rows of nunataks 
coalesced into 
mountain ranges; 
the ice cliff retreated 
up the bay, passing 
one nunatak after 
another and leaving 
it as an island or a 
promontory; one by 
one the tributary ice 
streams were aban- 
doned by the waning 
trunk and left as in- 
dependent glaciers, 
whose terminal ice 
cliffs retreated grad- 
ually up their several 
sea arms. These 
modifications of the 
coastal geography 
were accompanied by equally remarkable changes at 
higher levels. Several glaciers lost their neves, by dis- 
sipation or by diversion, and being thus deprived of 
nourishment and of part of their propelling force, lie 
nearly stagnant and are slowly melting away. A num- 
ber of fragments of ice streams are stranded on flat 
passes among the hills, where they lie almost inert but 


Ruled areas, land. BG, Brady Glacier. CG, Char- 
pentier Glacier. GI, Geikie Inlet. GP, Grand Pacific 
Glacier, HI, Hugh Miller Inlet. HM, Hugh Miller 
Glacier. JH, Johns Hopkins Glacier. MI, Muir Inlet 
MG, Muir Glacier. Q, Queen Inlet. R, Rendu Inlet. 
RI, Reid Inlet. W, Willoughby Island. 


with slow movement on two sides in opposite directions, 
part of the mass tending backward toward its original 

Of the abundant evidence from which this history was 
worked out by my predecessors ^ I saw only a small frac- 
tion, but enough to substantiate their conclusions in all 
essential respects. I skirted the barren coasts over which 
vegetation is slowly creeping from the south. I saw 
glacial till and gravel charged with bruised trunks and 
boughs from the ancient forest. I saw the bare ice-carved 
hills, still retaining striae and polish under a climate that 
has obliterated from most exposed surfaces the similar 
records of Pleistocene glaciation (pi. xviii). I saw the 
banks of stratified gravel before Muir Glacier — remnants 
of the old moraine-delta — and noted that their upper 
surface had been first sculptured by the readvancing 
glacier and then sheeted with till during the subsequent 
retreat. And I saw a remnant of the ice flood stranded 
on a saddle a thousand feet above tide. 

The saddle to which I refer is part of a small trough 
extending southeast from Hugh Miller Inlet and lying 
parallel to the adjacent great trough of Glacier Bay. 
During the recent ice maximum an ice current followed 
this trough from northwest to southeast, and when the 
supply from the northwest finally ceased and this strand 
of ice had nothing to urge it except its own weight, its 
ends slid into neighboring valleys, but the central part lay 
balanced on the summit and became stagnant (fig. 8). 
The adjacent hills are too low to furnish the snow needed 
to replenish its annual loss by melting, and so it is slowly 
wasting away 

1 Wright, Am. Jour. Sci., 3d series, vol. xxxiii, pp. 11-18, 1887; Ice Age, 
pp. 55-62, 18S9. Reid, Bull. Geol. Soc. Am., vol. 4, pp. 32-41, 1893; Sixteenth 
Ann. Rept. U. S. Geol. Survey, part i, pp. 434-442, 1896. Gushing, Am. Geol., 
vol. VIII, pp. 314-224, 1891. 



Of the existing glaciers of the basin ten are now tidal. 
We visited the Muir, Reid, and Hugh Miller. To the 
Grand Pacific, Johns Hopkins, and Charpentier we ap- 
proached so near as to obtain some information as to re- 


The ice mass occupies a typical glacial trough, shaped by an ice current flowing southeast- 
ward, or from the observer. Photographed June lo, 1899. 

cent changes. Of the Carroll, Rendu, Geikie, and Wood 
we had only distant views. 

Muir Glacier, — The Muir Glacier is not only the larg- 
est of the group but the most accessible. It has been 


Photographed June 9, 1899, from point on mountain spur at the east, a little lower than 
point E, fig. 10. 



visited by many ship-loads of tourists; it has received the 
chief attention of students of physical geography; and 
much is known of its recent history. Muir's notes and 
sketches record the approximate position of its sea cliff in 
1880, and photographic and instrumental determinations 
of its outline were made by Wright in 1886, and by Reid 


Showing position of front of Muir Glacier at various dates. Reid's survey stations are 
indicated each by a letter, dot and circle, with figures showing height in feet above mean tide. 
A and B were the extremities of his base line. 

in 1890 and 1892. During our visit the outline was 
mapped by Gannett, who tied his work to monuments left 


by Reid. The accompanying map (fig. lo) is copied from 
Reid, with the addition of Gannett's line showing the con- 
dition on June 10-12, 1899.^ 

From 1880 to 1886 the cliff retreated about 4,000 feet; 
from 1886 to 1890, 3,300 feet. Between 1890 and 1892 
there was an advance of 700 feet; between 1892 and 1899 
a retreat of about 1,900 feet, nearly the whole of the 
change occurring after 1894. The net retreat in nineteen 
years was 8,500 feet, or 1.6 miles. 

Although several of these surveys were quite accurate, 
I have stated their results in round numbers, because 
greater refinement would tend to mislead. The forward 
flow of the glacier probably varies little from day to day, 
but the breaking away of the cliff is quite irregular and 
the outline of the front continually changes. Besides this 
change in local detail there is a general change of more 
permanent character, to be described presently; and it is 
probable that an important seasonal oscillation interferes 
with the direct comparison of observations made at differ- 
ent times of year. The average annual retreat of the ice 
cliff for the four years preceding 1890 was 800 feet, and 
for the following two years the average advance was 350 
feet; but Reid noted in the summer of 1890 a retreat of 
500 feet in 43 days (July 26-September 7), equivalent 
to an annual rate of 4,000 feet. Again in 1892 he made 
two observations 31 days apart (July 19- August 19), 
and these show retreat at the rate of about 3,500 feet per 
annum; but the whole retreat in the seven following years 

iReid's map is pi. xcv« in i6th Ann. Rept., U. S. Geol. Surv., part i, 1892. 
Gannett's work was limited to the changed ice front and the resulting extension 
of the shore line. Field notes and photographs show that there were associated 
changes in the drainage of the gravel plains, but as these data are not sufficient 
for the correct delineation of the streams in 1899 I have left them as Reid repre- 
sented them. Some time before 1894 the northern branch of the western creek 
found a shorter course to the inlet, reaching tide about three-fourths of a mile 
nearer the glacier; and some time between 1894 and 1899 a similar change oc- 
curred on the east side of the inlet. 


was only 1,900 feet. It is therefore probable that the 
summer rate of retreat is much more rapid than the winter. 

While these considerations tend to qualify the figures 
deduced from uncorrected observations, they do not affect 
their general tenor. The application of a correction 
for seasonal oscillation would diminish by a few hun- 
dred feet the estimate for the total retreat from 1886 to 
1890, and increase by a similar amount the estimate for 
the total retreat from 1892 to 1899, but would leave un- 
changed the general result that the retreat in nineteen years 
has been more than a mile and a half, and that the general 
retreat suffered at least one interruption, a small advance 
occurring between 1890 and 1892. 

The general change of contour mentioned above is of 
peculiar interest because it was predicted by Reid. The 
middle of the glacier ends in deep water, a maximum 
sounding of 720 feet having been recorded, but at each 
side of the rock trough the ice rests on a bank of gravel 
rising above the water level. Observing that the surfaces 
of these gravel banks descend northward beneath the 
glacier, Reid inferred that with the progress of recession 
lanes of sea water would soon be admitted between the 
ice and the gravel; and having determined that the mar- 
ginal parts of the glacier advance very slowly as compared 
to the medial, he inferred that their cliffs would be carried 
back with relative rapidity by the attack of the warm sea 
water. As the map shows, these expectations have been 
fully realized. 

In bringing the record of the glacier down to the sum- 
mer of 1899 the preceding paragraphs practically close a 
division of its modern history, for a new epoch was intro- 
duced only three months after our visit. On the 12th of 
September the southern coast of Alaska was shaken by a 
severe earthquake, and other shocks followed at intervals. 
These greatly modified the condition of its tidal glaciers. 


and the Muir, at least, has not yet, after the lapse of three 
years, resumed its normal habit. Just what changes were 
made we do not know, because ordinary sources of infor- 
mation have been cut off and no student of glaciers has 
yet made an investigation. 

On July 29, 1900, O. H. Tittmann, Commissioner for 
the United States on the International Boundary, visited 
Glacier Bay on the tourist steamer ^ Queen.' In the lower 
part of the bay the main channel was so obstructed by 
floating ice that the commander of the vessel sought for 
an easier passage west of Willoughby Island, and finally 
desisted from the attempt to approach the glacier when 
opposite the mouth of Muir Inlet. Looking ahead it ap- 
peared to Tittmann that the whole of Muir Inlet was occu- 
pied by an ice pack, the ice being probably grounded and 
stationary. Instead of being able to steam as usual to the 
very front of the glacier, the vessel was turned back at a 
point about ten miles distant. In the summer of 1901 
tourist steamers were stopped by the pack at distances from 
the glacier ranging from five or six to eight or ten miles. 
In December of the same year a special trip was made by 
the ' City of Topeka,' for the purpose of forecasting the 
accessibility of the glacier for the excursion season of 
1902. The way was then found comparatively open, and 
the steamer approached within about a mile of the gla- 
cier; but in the following summer the nearest approach 
was to a point five or six miles distant. 

These facts seem to show that the earthquake shock, or 
shocks, not only set free a great quantity of ice, including 
bergs of unusual size, but left the glacier in such condi- 
tion that bergs were more easily detached in subsequent 
years. The determination of the nature of that condition 
will be of much interest to students of the physics of 
glaciers. Joint systems in rock have been plausibly as- 
cribed to the passage of earthquake waves, and it is easy 


to understand that the comparatively weak and equally 
brittle material of glaciers may be still more susceptible 
to rupture by sudden strain. As the tension cracks inci- 
dental to the flow of glaciers are quickly welded, except 
near the surface, it would appear probable that earthquake 
cracks in the terrestrial parts of a glacier have no perma- 
nent effect of importance, but the case may be materially 
different in the parts encroaching on the sea. 

As heretofore known, the frontal wasting of Muir 
Glacier has been chiefly by melting below the water-line, 
and the ordinary bergs, produced by the shearing off* of 
the overhanging upper portion, have been of moderate size, 
readily floating away. The greater bergs which stranded 
in the inlet after the earthquake may have been produced 
by cracks which divided the glacier from top to bottom. 

Reid Inlet, — Glacier Bay parts at its head into three 
branches (fig. 7). The westermost division, Reid Inlet, 
receives the Grand Pacific Glacier from the northwest, 
the Johns Hopkins from the west, and the Reid from the 
south, the three fronts circling in compact order about 
the western or northwestern end of the inlet. In 1899 
(as also in 1892) this was the region of most active berg 
formation, and on the day of our visit, June 12, the float- 
ing ice was packed so closely as to stop our progress — 
with a rowboat — and we succeeded in reaching only the 
Reid Glacier. With the use of a plane-table, a map was 
made of the lower end of Reid Glacier, and imperfect 
topographic sketches of the ends of the Johns Hopkins 
and Grand Pacific; and the data thus obtained were after- 
wards combined with the representations of the same dis- 
trict by Reid^ and the Canadian Boundary Commission to 
produce the sketch map in fig. 11.^ 

1 Sixteenth Ann. Rept. U. S. Geol. Survey, part i, pi. Lxxxvi, 1896. 
•There is some confusion as to names of glaciers about the upper part of 
Glacier Bay. Muir, the explorer of the region, gave manuscript names, some 


In the original exploration of Reid Inlet in 1879, Muir 
found it headed by a single great glacier whose ice cliff 
spanned the fiord from wall to wall, being interrupted only 
by a single boss of rock, half island, half nunatak. This 
glacier he named the Grand Pacific. When Reid in 1892 
found and mapped two great glaciers, besides a third of 
moderate size, Muir wondered and was perplexed, for it 
did not seem possible that he could have overlooked two 
tidal glaciers; and it was not until he revisited the inlet 
in 1899 that the mystery was solved. During the thirteen 
years which had intervened between exploration and sur- 

of which received newspaper publication. Reid, who made the first survej 
(1892), adopted Muir's names and added others, his nomenclature being first 
fully published in 1896 ( Sixteenth Ann. Rept. U. S. Geol. Survey ; text and 
map). The Canadian International Boundary Commission, surveying the same 
region in 1894, prepared maps with a different nomenclature. These maps, be- 
ing primarily for the use of commissioners in connection with the pending 
boundary question, have not been ofl[icially published, but they have been unoffi- 
cially distributed and their data compiled in various maps. 

The following are the discrepancies : 

Reid^s name. Can. B. Com. name* 
Glacier reaching Reid Inlet from 

the north. Grand Pacific. Johns Hopkins. 

Glacier reaching Reid Inlet from 

the west. Johns Hopkins. No name. 

Glacier reaching Rendu Inlet. Rendu. Charpentier. 

Glacier reaching Hugh Miller 

Inlet from the south. Charpentier. No name. 

Glacier reaching Queen Inlet. Carroll. Woods. 

Glacier reaching Geikie Inlet 

from the south. Wood. No name. 

It would appear that the cartographer of the Commission had moved three 
names — Johns Hopkins, Charpentier, Wood(s) — from the glaciers to which 
they were attached by Reid, and given them to other glaciers, displacing the 
names given to these others by Reid. No new names are added, and the dis- 
placed names do not appear elsewhere. 

Reid's nomenclature is followed by the U. S. Coast Survey in chart 3095, 
Glacier Bay (1899). The Canadian Commission is followed by the U. S. Coast 
Survey in chart 8001, Northwestern coast of North America (edition of 1898), 
and in chart 3091, Territory of Alaska, southeast section (1898). Otto J. Klotz, 
an officer of the Canadian Survey, in the Geographical Journal (vol. xiv, 1899), 
uses Reid's name, Grand Pacific, in his text (p. 529), but in the titles to two 
figures (pp. 527, 529) applies the Commission name Johns Hopkins to the 
Grand Pacific Glacier of Reid. 

Reid's names have recently been adopted by the U. S. Board on Geographic 
Names. See National Geographic Magazine, vol. xii, page 87, 1902. 



vey the main trunk of the Grand Pacific had disappeared, 
leaving three of its branches as independent tidal glaciers. 
The name has been retained for the northwestern branch. 
In 1894 the district was resurveyed by the Canadian 
Boundary Commis- 
sion, and another 
record was made of 
the progressive 
changes of the gla- 

The Grand Pa- 
cific in 1892 and 
1894, as in 1879, 
presented two 
fronts to the waters 
of the inlet, the 
fronts being sepa- 
rated by a high rock 
island, but in the 
later years a much 
larger part of the 
island was laid bare. In 1899 — as nearly as my distant 
views enabled me to determine — the western and greater 
arm of the inlet had eaten back into the glacier so as to 
reach some distance beyond the head of the island and 
approach the mainland at the northeast, thus isolating a 
body of ice lying between the island and the mainland. 
The arrangement of moraines delineated by the Boundary 
Commission showed that in 1894 this body had ceased to 
be replenished by the current of the Grand Pacific, so that 
its complete wasting was a mere question of time.^ It is 
possible that in 1899 it had become so far reduced as no 
longer to touch the island. 


Showing positions of ice front in different years. 
I^and areas are ruled. 

^ These moraines are beautifully shown in one of the Commission's photo- 
graphs : A. J. Brabazon, No. 45, vol. 14, p. 13. 



The front of Johns Hopkins Glacier has suffered less 
change since Reid mapped it, but there has been some 
retreat. A comparison of photographs made in 1892 and 
1899 shows that more rock is exposed at the south; and 
a remnant of ice I could see clinging to the mountain side 
at the north showed that the front had within a very few 
years held a position several hundred yards more advanced 
than in 1899. Its determined changes were so small that 
no attempt has been made to express them in the accom- 
panying diagram. 

The remaining glacier of the three was indicated in a 
general way on Reid's map of 1892, but its lower end was 
represented by a dotted line, implying doubt as to its pre- 
cise position, and no name was attached. It was more con- 
fidently delineated by the Boundary Commission in 1894. 
As my map and photographs give such determination of 
the position of its ice cliff that future changes can be meas- 
ured, it seemed proper to supply Reid's omission in re- 
spect to name, and the Harriman Expedition adopted 

the name Reid. I 

am glad to be able to 

^^E illustrate somewhat 

fully a feature bear- 

ing a name so de- 
serving of honor in 
Alaska glaciology. 
As already men- 
tioned, the Reid Gla- 
cier was a branch of 
the Grand Pacific in 
1879, but not many 
years could have 
elapsed before the recession of the latter made it indepen- 
dent, and it is probable that its end then projected some- 
what beyond the general line of the south wall of the 

FIG. 12. REID glacier; DISTANT VIEW 


From a photograph by A. J. Brabazon. 



inlet. One of Reid's photographs shows that in 1892 
it had retired within the general line of the mountain 
front. The Boundary Commission's map places it outside 
the position of Reid's dotted line; and in 1899 it was a 
half mile within the limiting capes. A comparison of the 
Commission's photographs^ with my own indicates a re- 
cession of about 1,500 feet in five years, accompanied by 
an important modification of the character of the front. 
The removed portion was part of an ice cascade, and 
as there was little 
change in the thick- 
ness of the glacier, 
the terminal cliff in 
1894 was much 
lower than in 

I found three 
masses of dead ice, 
testifying to the 
former greater ex- 
tent of the glacier. 
One of these masses 
rested on a small 
promontory just east 
of the glacier front. 
A knob of marble is connected with the main mountain by 
a comparatively low saddle, and on this saddle lay a body 
of ice apparently several scores of feet in depth but almost 
wholly concealed by stones and gravel. This was outside 
the line of flow of Reid Glacier, and could only be a 
remnant of the mass of the Grand Pacific when that 
glacier occupied the full width of the inlet. It dated back 
to a time when the conditions were as observed by Muir 
in 1879, and its gradual melting had probably been in 

* Especially A. J. Brabazon's No. 38, vol. 14, p. 11. 

From a photograph by A. J. Brabazon. 



« S 



V a 

M U 

^ 'a 

ON o 

00 6 

p. ^ 

2. =i 




a -3 '5 

2 S 


•H ** 
u w 
I' ^ 

"o 9 

li o 


« « 
a s 

g ^ 

O M s 

4> O 3 
V- J3 « 

<u S * 
' . ''3 is 

t; ?« fl 

o ^ « 

5 S -^ 
g w - 

" 01 S3 

• a .2 a 

i-o bo ^ 

C V o 

£5 a 




progress for nearly twenty years. The other masses clung 
to the walls of the trough occupied by Reid Glacier, and 
were still continuous with the ice of the glacier, although 
they had ceased to move. They were simply portions of 
the retreating glacier so well supported by the land that 
they had not fallen into the sea when the deeper parts of 
the ice stream were melted. They were partly covered 
by gravel and other rock debris, but still showed faces of 
white ice near their junction with the glacier. That on 
the west side extended about 1,200 feet beyond the general 
ice front, and the eastern mass was nearly or quite as long. 
From these remnants it is inferred that there was a some- 
what gradual shrinkage of the glacier after the record of 
1894, and that it did not immediately assume the propor- 
tions observed in 1899. 

The width of the Reid at its debouchure is seven-eighths 
of a mile, and the general height of its ice cliff about 100 
feet. The frequent falling of ice masses during our visit 
gave the impression that it was discharging bergs rapidly, 
but the little bay before it carried less floating ice than the 
open inlet beyond. 

The data compiled in the map of the inlet (fig. 11) are 
of unequal precision. The line representing the ice front 
in 1879 depends largely on the recollection of Muir, 
but his recollection is supported by notes and a landscape 
sketch made at the time. The fronts of Grand Pacific 
Glacier in 1892 and 1894 are believed to be close approxi- 
mations, but the front in 1899 has much less authority and 
may involve considerable error. 

The total retreat along the axis of Johns Hopkins Glacier 
in the twenty years preceding 1899 was about three and 
a half miles, and the retreat of the Grand Pacific along 
the line of its western distributary was three and a half 
to four miles in the same period. Reid Glacier retreated 
a half mile after its separation, the period being some- 



thing less than twenty years. In 1879 the Grand Pacific 
had a total water front of over three miles; in 1892 and 
1894 the separated glaciers presented a total front of more 
than six miles, the exposure to the sea being progressively 
increased up to that time. Afterward the length of front 
underwent little change, and should recession continue 
it will diminish. Unless the enlargement of the ice front 
was compensated by shoaling of the water, the sea had 
exceptional advantage for twenty years in its attack upon 
the ice, and this advantage may have been connected with 
the phenomenal rate of recession. 

Bergs, — The upper part of Reid Inlet in 1899 not only 
contained more floating ice than any other portion of 
Glacier Bay, but its bergs were of greater size than any 
others we saw. The one pictured in fig. 16 was ascer- 





Scale is given by a boat. The nearer face of the berg, black with embedded detritus, is 
probably part of the base of the parent glacier. Photographed June, 1899. 

tained to rise 76 feet above the water, and one of my 
companions who landed upon it and walked from end to 
end, estimated its length by pacing at 750 feet. Having 
my eye gaged by these measurements, I was able to form 
an approximate judgment of the size of other large bergs 
seen; and I estimated the largest, which was approxi- 
mately tabular and square in form, to measure 1,000 feet 
on each side and to rise 100 feet from the water. The 
thickness was probably not less than 500 feet. These 



dimensions raised questions as to the mode of calving. 
Reid, from studies of the Muir, inferred that waste along 
the ice front was chiefly from melting by sea water; and 
this melting, in combination with the forward flow of the 
ice, produced an overhang, resulting in the breaking away 
of the upper portion of the glacier by its own weight. 
Such observations as I was able to make at various ice 
clifls were in full accord with his view, and the great 
bergs of Reid Inlet may perhaps have been formed in that 
way; but their size led me to wonder whether another 
process might not be involved. Many of the immense 
bergs of the Antarctic Ocean are tabular in form, and it is 
believed that they have the full thickness of the parent 
glaciers, which were protruded into the sea until actually 
floating upon its surface before the separation took place. 

In order that the Grand 
Pacific or the Johns Hop- 
kins should produce 
bergs of this type, it 
would be necessary that 
the depth of the fiord in 
front of the ice cliff" 
should be not less than 
seven-eighths the thick- 
ness of the glacier. 

Hugh Miller Inlet, — 
Hugh Miller Inlet oc- 
cupies an irregular re- 
cess among the hills and 
mountains on the south- 
west side of Glacier Bay. 
Rocky islands of moder- 
ate elevation half block its entrance and interrupt its sur- 
face. Charpentier Glacier reaches it from the south, ter- 
minating in a low ice cliff" about seven-eighths of a mile 


Showing positions of the ice front in different 
years. I^and areas are ruled. 



broad. The Hugh Miller, descending from the moun- 
tains at the southwest, spreads into a broad field north- 
west of it, and has a double discharge, one part coming 
eastward to the inlet, and the other going northward 
toward Glacier Bay. The face toward the inlet is about 
three and a half miles long. The southern third overlooks 
the water in a low cliff, and the remainder presents a 
sloping surface black with accumulated rock debris. 
Though the land-locked inlet gives the floating ice scant 
opportunity to escape to the open bay, very little was ac- 
cumulated at the time of our visit, June ii, and all the 
bergs were small. 


Photographed in 1899. The till is thin, but contains remnants of ice, and was englacial in 
part. The single bush visible, almost the only vegetation on the new ground, is a willow. 

The dates of exploration and survey are the same as for 
Reid Inlet and the general history of change is strictly 
parallel. Muir observed but a single glacier, to which he 
gave the name Hugh Miller. Reid found two, and added 
the name Charpentier. With the aid of Reid's map I was 
able to indicate somewhat definitely the extent of the sub- 
sequent recession, and Muir, in revisiting the locality with 



me, was able to say that the change previous to Reid's survey 
was greater in amount than the more recent modification. 

But after the lapse of twenty years Muir found it im- 
possible to recall the precise position of the ice front in 
1879, and a subsequent study of his notes and sketches 
left the matter still in doubt. On the accompanying map 
(fig. 17) I have drawn two tentative lines, one crossing the 
inlet near its mouth, the other running northward from 
the peninsula which margins the Charpentier Glacier. 
The more conservative of these implies that the face 
of Hugh Miller Glacier retreated about one and a half 
miles between 1879 ^^^ 1892. 

In 1892 Charpentier Glacier had two fronts, separated 
by its contact with a broad rock island or peninsula. One 
front, facing northward, was tidal; the other, facing north- 
eastward, was non-tidal, but ended close by the water. 


Its surface was dark from the accumulation of rock debris, 
and its motion had probably ceased. Seven years later 
the tide-water front had retreated about one-third of a 
mile, thereby nearly severing its connection with the inert 
mass at the northeast (Z>,fig. 19), and the waste of the latter 


had progressed so far as to render its surface almost black 
from the accumulation of residuary moraine stuff. These 
relations are indicated in figure 17. The condition in 
1894 was intermediate between those of 1892 and 1899. 
A comparison of Reid's photographs and mine shows 
also that the Charpentier had lost in thickness as well 
as area, the lowering near its front being estimated at 
about fifty feet. 

As mapped and described by Reid, Hugh Miller Glacier 
rested at one point against an island, and only the portion 
south of the island yielded bergs. The northern division 
of the front descended to tide-water, but was covered at 
its margin by debris and had no cliff. In 1899 the front 
had retreated so as to open a narrow channel west of the 
island and expose the top of another island on which a 
tongue of the ice rested. South of this point the ice cliff 
had retreated westward to an average distance of 1,000 
feet, the maximum being nearly 2,000 feet. It still 
yielded bergs, but sparingly, and was probably approach- 
ing a non-tidal condition. A large nunatak at the south, 
which was mapped and photographed by Reid, was more 
fully exposed than before, and a small one had appeared 
between it and the ice cliff. Near the northern end of 
the cliff the distortion of dirt bands and an uprising of the 
surface of the ice suggested that another rocky islet would 
soon be exposed. The retreat of the northern portion of 
the glacier face had laid bare a group of rocks projecting 
slightly above the water, and a larger rock knoll was 
gradually emerging. At one point it projected as a nun- 
atak about 150 feet above water-level, and farther on was 
revealed at the water's edge. The average width of the 
space here abandoned by the ice is about 1,500 feet. 
(See pi. III.) The condition in 1894, as indicated by 
photographs, was intermediate, but nearer to the condition 
in 1899 than to that in 1892. 


At the northern margin of the Hugh Miller ice field, 
where it discharges toward Glacier Bay, there is some- 
what similar evidence of retreat. Muir's sketch shows 
that this front reached tide-water in 1879, but was inter- 
rupted near its eastern margin by a small island. This 
detail serves to fix its approximate position as represented 
in fig. 17. Reid states that at the time of his survey it 
was non-tidal, but his map places the ice margin at the 
water's edge. Photographs made in 1894 indicate a bare 
tract 1,500 or 2,000 feet broad between the ice front and 
the strand, and at the time of my visit the distance was 
2,600 feet, the surface being chiefly occupied by ground 
moraine. The ice front in 1899 had a gradual slope, was 
covered by drift near its margin, and was traversed by a 
large medial moraine. Close to its front it received a 
tributary, cascading down a narrow valley from the west. 
The total retreat of the ice front at this point was prob- 
ably a little less than one mile in twenty years. 

Geikie Inlet, — Exploring Geikie Inlet in 1879, Muir 
found it headed by a tidal glacier, to which he gave 
the name Geikie. His notes estimate its width as several 
miles, but do not serve to fix the position of its front. In 
1892 Reid found that its front had receded so far as to 
convert its two branches into distinct glaciers. Retaining 
the name Geikie for the more northerly, he called the 
other Wood Glacier. The Geikie was tidal; the Wood 
barely touched the water at two points, but yielded no 
bergs; and the nearer corners of the glaciers were con- 
nected by a short body of motionless ice. Photographs 
made by the Canadian Boundary Commission in 1894^ 
show that in the two years elapsed since Reid's survey 
both glaciers had shrunk, the Wood receding several 
hundred feet and the Geikie about half a mile. The 

* A. J. Brabazon, nos. 32, 38, 39 and 40, contained in volume 14 of the official 
album, pp. 9, 32 and 33. 



Geikie was still tidal. The two were still connected by 
stagnant ice, a long, narrow strip, partly protected from 
melting by moraine stuff. In 1899 we failed to reach the 
head of the inlet, so that my only contribution to its his- 
tory consists in assembling the observations of others. 


Fairweather Range, which bounds Glacier Bay on the 
southwest, presents its other face to the open ocean. Its 
seaward face is bold and lofty, and the greater part of it is 
above the snow-line. While rugged in detail it is little 
complicated by foothill ranges, and from the ship's deck 
we could trace a number of its long glaciers from end to 
end. Between its base and the sea there is usually a 
narrow foreland, but this disappears toward the east and 
broadens toward the west. At several points it gathers 
the alpine glaciers into massive piedmont sheets, and 
several of these approach or actually touch the sea. The 
La Perouse, at whose edge we made a landing, is of this 
type, being fed by alpine glaciers from the slopes of the 
range about Mounts Crillon, D'Agelet and La Perouse. 


Showing the alpine glaciers that feed the plateau mass below ; and the relation of the 
timbered ridge at the left. 

Its extent in the direction of the coast is about three 
miles, the central portion ending in a lofty white cliff. 
The eastern third is darkened by a covering of moraine 
and appears to be separated from the water by a strand. 



The western part is partly concealed by a timbered ridge 
running parallel to the coast (fig. 20). This ridge, which 
is probably a huge moraine of Pleistocene age, extends 
westward far beyond the end of the glacier. Our landing 
(June 18) was at the western end of the ice cliff, coincid- 
ing with the eastern limit of the timbered ridge. 

The ice cliif facing the ocean at this point has the gen- 
eral appearance of the front of Muir Glacier, but its sub- 
merged profile is different. Instead of deep water it 
overlooks a shoal, from which boulders project here and 
there and on which we saw small bergs stranded.^ Very 
little floating ice was visible, and no large bergs. The 
cliff is evidently sapped at base by the wash of the waves, 
and the process which perpetuates it is closely similar to 

the process which main- 

tains rock cliffs along other 
portions of the coast. 

The glacier, which farther 
west presses against the 
timbered ridge, flows past 
its end to the sea, and thus 
the extremity of the ridge 
occupies a reentrant angle 
in the margin of the glacier. 
Close to the angle a stream 
of water, escaping from the 
glacier, has crossed the 
ridge, eroding a deep gash, 
in whose walls the structure 

5j?» «'<4'»\ ♦'% 




The row of circles marks the position of 
a fresh-formed moraine ridge, with overturned 
trees. Forest is indicated by stars, sand beach 
by dots. Approximate scale: 1 inch =2,000 feet. 

of the ridge is revealed. The walls are not clothed by 
vegetation, but are somewhat cumbered by the trunks of 
forest trees fresh-fallen from the crests on either side. The 
stream has been very active within the last decade or two, 
and it seems probable that all its work of erosion was 

* If our visit was at high tide, this shoal may have been bare at low tide. 



performed within that period. The section it exposes 

(fig. 22) includes horizontal beds of blue clay, flanked on 

the seaward side by highly inclined beds of similar clay 

alternating with layers of sand. Both these are truncated 

above, and 

overlain un- 


by a series of 


bedded grav- 

^ . V. * "U ^' ^^*^^"^'^^ ^^^y- ^> bedded clay and sand. C, bedded gravel, 

els, in WniCn with angular boulders and trunks of trees. D, bouldery till, with 

are incorpo- *-^*^-^*- ^'^^--^• 

rated large angular boulders and trunks of trees. This 

gravel is succeeded on the landward side by — and prob- 

FIG. 22. 



The glacier is at the right. Between it and the forest is a tract occupied by fresh drift, 
with sticks and logs. Photographed in June, 1899. 

ably passes into — a bouldery till, which also contains 
trunks of trees. The clays and sands evidently represent 
an epoch when the coast was more deeply submerged than 



now, but our meager facts include nothing to indicate either 
the date of their deposition or the date and manner of their 
deformation. The overlying gravel and till are clearly of 
glacial origin, and the gravel was laid down at or near sea- 
level. As it is now about 150 feet above the sea, it is evident 
that there has been a change here in the relation of land 
and ocean. The buried tree trunks tell of an advance of 
the glacier over a tract that had existed as dry land. 


The glacier is out of sight at the left (compare fig. 23). The moraine, here 10 feet high, is 
crowded against forest trees, and includes crushed trees. Photographed in June, 1899. 

The remnant of timber standing east of the stream val- 
ley was separated from the glacier at the time of our visit, 
by a belt of barren ground from loo to 200 yards wide (fig. 
23). This ground was occupied by bouldery till containing 
bruised and macerated branches and trunk fragments, and 
the margin of the timber showed unmistakable evidence 
of recent attack by the ice (fig. 24). The till had been 



pushed up into the forest, forming a heap several yards 
in height, and stones and earth were mingled with trunks 
of trees and other vegetal debris. It was evident that the 
forest had recently extended somewhat down the slope 
toward the present position of the ice, and that a temporary 
enlargement of the ice field had crowded it back. This 
had occurred so recently that a younger growth of trees 


had not yet started on the morainic ridge. Some of the 
overthrown trunks still retained their bark, though it had 
fallen away from most. The wood of the trunks was 
still sound, but some branches an inch and more in diam- 
eter had become brittle, and leaves and smaller twigs had 
fallen off. With the local woodman's knowledge of the 
rate of decay in an Alaska forest it would be possible to 
estimate closely the date of the advance. To my inex- 
pert judgment it appeared probable that it occurred within 
the last decade of the century. 


Since writing the last paragraph I have received from 
the U. S. Fish Commission a photograph made from the 
steamer ^Albatross' in September, 1895, and have repro- 
duced it in figure 25. This shows that the ice at that 
time was in actual contact with the forest. 

In the standing timber, which comprises both spruce 
and hemlock, and also among the overturned trees, are 
trunks four feet in diameter, and during the whole life of 
these trees — a matter of centuries — this particular spot 
has been undisturbed by the ice. It is thus shown that 
glaciation has here attained within a few years a maximum 
not previously reached for centuries. 

The locality of our visit records at least two glacial 
maxima. As the older till contains tree trunks, it marks 
an advance of the ice after an epoch of inferior develop- 
ment. The mature forest standing on this till records 
another long epoch of lessened glaciation, and the recent 
advance a second maximum. It is possible that the epoch 
between the two maxima was of only a few centuries, but 
evidence to be mentioned in another connection indicates 
that it was much longer. 

On our return voyage in July we passed this part of the 
coast at a distance of several miles, and I was able to note 
that the eastern margin of the piedmont division of La 
Perouse Glacier lay parallel to a forest margin, with an 
intervening belt of different color, presumably morainic, 
about 200 yards wide. Similar belts were also seen about 
the flanks of the next piedmont mass toward the east. 
These facts indicate that the recent advance recorded at 
one point on the front of La Perouse Glacier was an ad- 
vance of the whole glacier, and render it probable that 
the change was not confined to a single glacier. 

The evidence from the crushing of the forest does not 
tell us whether during the long period before the last ad- 
vance the ice had approximately its present extent or ex- 


perienced an important minimum, but the latter history is 
rendered probable by comparison with facts recently de- 
veloped about other glaciers of the same mountain face. 
Lituya Bay, fourteen miles northwest of La Perouse 
Glacier, was explored in 1786 by La Perouse, who de- 
scribed and mapped the principal glaciers descending to 
it. Klotz has made a comparison of La Perouse's account 
with the condition found by himself in 1894 (fig. 74), 
and shown that there has been a marked advance of the 
ice in both arms of the bay, the western glacier encroach- 
ing three miles on the water of the bay and the eastern 
two and one-half miles.^ The foot of Brady Glacier, 
twenty-five miles east of the La Perouse, was visited by 
Vancouver in 1794, and from a discussion of his descrip- 
tion Klotz concludes that the ice front was then at least 
five miles less advanced than in 1894. Muir in 1880 
found the margin of the Brady against a mature forest 
whose territory it was invading. As the La Perouse lies 
between glaciers of the same range which have experi- 
enced a great advance, and as it has recently crowded 
against a forest, the probability is that its history resembles 
that of its neighbors and includes a great forward move- 
ment during the last century. 


In the neighborhood of Yakutat Bay a foreland fifteen 
to twenty-five miles broad separates the mountains from 
the open ocean. The bay lies partly in the foreland and 
partly among the mountains. The outer part, to which 
the name Yakutat is more specifically applied, is nearly 
twenty miles broad, but narrows toward the mountains. 
The inner part penetrates the mountain district for ten 
miles in a north-northeast direction, with an average width 

^ Notes on Glaciers of southeastern Alaska and adjoining territory. Geog. 
Jour., vol. XIV, pp. 523-534' 1899. 



of three miles, and then turns to the right at a sharp angle, 
and, assuming the character of a narrow fiord, runs back 
thirty miles toward the south-southeast (fig. 26). It then 

passes from 


to the 


Based on map of the Canadian Boundary Commission, with local 
details by I. C. Russell and Henry Gannett. 

foreland, and 
ends in an oval 
three miles 
wide. The 
shorter and 
broader reach 
within the 
mountains is 
called Dis- 
Bay ; the long, 
narrow arm, 
Russell Fiord. 
Russell Fiord 
has two east- 
ward branch- 
es. The north- 

ern is at the present time about eight miles long, ending 
at Nunatak Glacier, and it is convenient to call it Nunatak 
Fiord. The southern, about one mile in length, leads 
toward Hidden Glacier. 

The foreland southeast of the bay is in general low, but 
includes hills and ridges of morainic aspect. So far as 
known, it is wholly constituted of till and gravel, brought 
by glaciers and associated streams of water when the ice 
fields of the region were more extensive. Northwest of 
the bay the foreland is occupied (or constituted) chiefly 
by a great piedmont glacier, the Malaspina, the ice being 


separated from the water of the bay by a belt of detrital 
lowland. The mountain system is lofty, and among its 
summits are great tracts of neve. From these a series of 
alpine glaciers stream down to feed the Malaspina, and 
others reach or approach the land-locked arms of Yakutat 
Bay. Turner Glacier, entering Disenchantment Bay from 
the northwest, flares at the end after the manner of the 
Davidson, but has not yet surrounded itself by a moraine 
barrier, and ends in a berg-producing cliif. The Hubbard, 
coming in two principal streams from the north and with 
minor affluents from the east, reaches the sea at the junc- 
tion of Disenchantment Bay with Russell Fiord and 
occupies the coast for more than five miles. Nunatak 
Glacier flows northwestward to the end of Nunatak Fiord, 
where it maintains a discharging cliff* nearly a mile broad. 
Hidden Glacier, with branches from the east and south, 
follows a trough parallel to Nunatak Fiord, but fails to 
reach tide-water, being separated from it by a gravel plain 
two miles long. 

Two islands should be mentioned here, not as important 
geographic features but as landmarks to which the follow- 
ing pages make occasional reference (see fig. 27 and pi. 
viii). Haenke Island, a rounded rock knoll several hun- 
dred feet high, lies near the east shore of Disenchantment 
Bay. Osier Island, lower but containing also a nucleus 
of rock, stands at the entrance to Russell Fiord. 

The inner arms of the bay were explored by Russell in 
1890 and 1 891, and he prepared a sketch map showing 
the general relations of fiords and glaciers. The shores 
were afterward surveyed by the Canadian Boundary Com- 
mission (1895), and during our visit Gannett made local 
maps of the upper part of Disenchantment Bay and the 
ends of Nunatak and Hidden glaciers. 

The prevailing rocks are friable sandstones and partially 
altered shales, and these are weathered and eroded much 


more readily than the rocks about Glacier Bay. For this 
reason the finer details of ice sculpture are preserved only 
on surfaces from which the glaciers have somewhat re- 
cently retreated. It is probably because of the rapid 
weathering that vegetation occupies ice-freed surfaces 
rather quickly, but this remark applies only to herba- 
ceous plants and to willows and alders having the habit 
of bushes. The spruces, whose dense forests cover the 
detrital foreland east of the bay and climb the seaward 
slope of the adjacent mountain, have accomplished little 
toward the invasion of the most freshly glaciated faces of 
the same mountain which are turned toward Yakutat Bay 
and Russell Fiord. 

So far as known by direct observation, the recent glacial 
history is one of waning and retreat. From a careful 
compilation of early records, made by Russell, it appears 
that Malaspina in 1792 and Vancouver in 1794, attempting 
to penetrate Disenchantment Bay in boats, found a glacier 
front at Haenke Island. This was essentially the face of 
Hubbard Glacier, to which the Turner was then trib- 
utary. Completely filling the head of Disenchantment 
Bay, it acted as a dam separating Russell Fiord from 
Yakutat Bay, and the fiord was then occupied by a lake. 
The discharge of the lake must have been southward over 
the gravel lowland, and during its existence the wash of 
its waves produced beaches which are still to be seen as 
terraces about the southern part of the fiord. Russell 
estimates their height above tide- water at 150 feet. 

From 1792 to 1899 the face of Hubbard Glacier re- 
treated about five miles, but there is no reason to suppose 
that its position in the days of Malaspina represented a 
maximum. Haenke Island, which was not wholly cov- 
ered by the ice at the time of Malaspina's visit, neverthe- 
less preserves glacial striation over the whole of its crest, 
and in places even polish; and this could hardly be the 


case if it had been exposed to the weather for more than 
one or two centuries. Moreover, there is no change in 
the vegetation at this point. The alder thickets which 
begin at the head of Disenchantment Bay, characterize 
the slopes of the mainland not only to Haenke Island but 
for miles beyond, and the first spruces noted were not less 
than five miles to the south of the island. But while the 
ice seems to have recorded neither a maximum nor a pro- 
longed lingering at the point where it was earliest observed, 
our present data suggest no other line of critical impor- 
tance. We can only say that for a period considerably 
greater than a century the general character of ice change 
has been diminution. 

Since the last paragraph was written, the U. S. Coast 
and Geodetic Survey has published a new chart of Yaku- 
tat Bay, giving soundings from the ocean to Point La- 
touche, six miles below Haenke Island. These sound- 
ings give no indication of a moraine in the vicinity of Point 
Latouche. Not far from that point there is a depth of 
i,ooo feet, and thence southward the channel is shown for 
five miles. Here, at a distance of twelve miles from 
Haenke Island, is a submerged bar with a depth of about 
300 feet, and this is probably the last-formed important 
moraine in the bay. There appears to be another oppo- 
site Knight Island, seventeen miles from Haenke Island, 
the intervening hollow having an extreme depth of about 
600 feet (see fig. 27). 

This is the greatest distance to which the channel of 
the Hubbard, or Disenchantment Bay Glacier can be 
clearly distinguished. Its course is not central to the bay 
but nearer the eastern shore, the Disenchantment Bay 
stream apparently having been crowded over by the ex- 
pansion of the Malaspina Glacier, which then included the 
Lucia. Farther to the south and southwest the soundings 
reveal a series of troughs and ridges whose trend and 



curvature suggest that they are connected with earlier 
positions of the margin of Malaspina Glacier, and these 
corrugations end against a great shoal which spans the 
mouth of the bay from cape to cape. This shoal is 
slightly convex toward the ocean, and its submergence 
ranges from 50 to 80 feet. It is believed to mark a former 
extension of the seaward face of Malaspina Glacier. 

Adapted from U. S. Coast Survey. Places where no soundings were made are indicated by ??. 

All or nearly all of these ridges are probably connected 
with the Pleistocene history of the region, but the moraine 


midway between Point Latouche and Knight Island may 
correspond to a comparatively recent maximum of the 
Disenchantment Bay Glacier — that from which it was 
retreating when observed by La Perouse. A more intel- 
ligent judgment can be formed when the system of sound- 
ings shall have been carried through Disenchantment Bay. 

Turning attention to Russell Fiord, we find pertinent 
phenomena somewhat more abundant, and it seems pos- 
sible that their careful study may yield important chapters 
of the local history. The lower parts of the fiord walls 
are finely sculptured, showing by magnificent flutings that 
there has been much longitudinal scouring. At various 
points, but especially south of Hidden Glacier, there are 
marginal banks of gravel similar to those about Muir 
Glacier, characterized by horizontal bedding but showing 
by their surficial forms that they have been overridden 
and molded by a glacier. Nearly all parts of the walls of 
Russell Fiord carry vegetation, of which the alder is a 
conspicuous element, but the growth is relatively sparse 
toward the north and dense and luxuriant toward the 
south. The gravels about the basin at the extreme south- 
ern end bear a luxuriant and mature forest of spruce. 

Looking back to the time when Russell Fiord was filled 
by a glacier, it seems evident that the ice stood for a con- 
siderable period with a front just outside the mountains. 
The expansion of the fiord in the edge of the foreland 
corresponds, I conceive, to the flaring end of David- 
son Glacier, and the surrounding plain of gravel is the 
equivalent of the moraine barrier which the Davidson has 
built in Lynn Canal. This condition is assumed to date 
back several centuries, for it is not probable that the 
forest could occupy the whole surface of the gravels until 
the ice had retreated. The banks of gravel within the 
fiord record lingerings of the ice front and subsequent re- 
advances, but whether these oscillations preceded or fol- 


lowed the epoch when the entire fiord was filled with ice 
is a matter of doubt 

Coming to this region while the features of Glacier Bay 
were fresh in mind, I searched gravel and till, wherever 
opportunity offered, for vestiges of earlier forests which 
might have been overridden by the glaciers, but the search 
was unsuccessful; and so far as my evidence goes, the 
condition of expanded glaciers observed by Malaspina 
and Vancouver may not have been preceded in this 
locality and in recent geologic times by a condition of 
relatively contracted glaciers such as now obtains. Rus- 
sell found a buried forest under the foreland gravels at 
the south end of Russell Fiord,-^ but the demonstrated 
oscillation need not have been of great extent. 

Hidden Glacier, — The valley in which the distal part 
of Hidden Glacier lies is a characteristic glacial trough 
with rather uniform cross-section. Its course curves from 
northwest to a little south of west, and it joins Russell 
Fiord at right angles. In 1899 for a distance of a mile 
and a half it was occupied by tide-water with a width of 
three-quarters of a mile; then came a tract of gravelly 
alluvium, nearly two miles long and a little narrower than 
the inlet. The glacier itself had a width of a mile. The 
ice front sloped gradually down to the alluvial plain, and 
although the profile was slightly arched, its greatest de- 
clivity (measured on a photograph) was only ten degrees. 
In the lower mile the surface was remarkably smooth and 
there were no important crevasses. There were lateral 
moraines, and near the southern of these a single strong 
medial, but the general face was exceptionally free from 
drift. Close to the front margin the ice was somewhat 
discolored, but so nearly white as to suggest that the 
lower layers, usually dark with englacial drift, were not 
visible. That they really lay at some distance below the 

1 Thirteenth Ann. Rept. U. S. Geol. Survey, Part ii, p. 89, 1893. 

-rbfiBid . !;IO nohhfi 

3fh fli I) ibrJHlJ^ fflOI' 

tfi'ji iu\^ ^llfitmiiO TinoH v/f bov/J/ni' 

-31 o1 ^ir <) '>^ornfK» !«f'>')rr .,j>^ 'jflj bin: "laiji:! 

bin, n 'Ji\j in fHiiob'iHoq orlj 

I ) V(f univ/inb 'j'ji'fi 



Map of Hidden Glacier 

Hidden Glacier reaches the sea at the head of a short arm, branch- 
ing from Russell Fiord. The whole of the arm is included in the jiri 
map. For general relations of the locality, see the map of Yakutat ^.1] 
Bay, figure 36. For description of the glacier see pages 52—58. 

Surveyed by Henry Gannett, June 20, 1899. The chief instrument 
used was a plane-table, and all stations were on the delta plain between 
the glacier and the sea. The special purpose of the work was to re- 
cord the positions, at the time of survey, of the ends of the glacier and 

Office drawing by Gilbert Thompson. 

markably smooth 




6 . 






« ?5 ^ 

>t-i <-■ ^ 
° ts s 

i2 2 ja 
. .2 .9 

V U P4 

P tn «, 


o u 

5 S fl 

11 bo +j 

4^ « ^ 


^ ja > 

bo u _, 

-2 as 

* ti ii 
« 5 « 


co . 

« ;S « 

s . ba 

CO n ^ 
« ^ -^ 
•2 •" >» 


5 <-* «J — 

4> 41 * Ol 

^ -^ a ^ 
"c -"^ 2 


visible glacier snout and were extended along the floor of 
the valley beneath the gravel plain, seemed to be shown 
by other phenomena. First, the water of ablation, though 
flowing out from the surface of the glacier near each side, 
did not visibly escape from the ice in the central part, but 
oozed up through the contiguous gravels, gathering in a 
number of streams, which attained their full size about a 
quarter of a mile from the ice front. Second, the gravel 
plain in the vicinity of the ice front was dotted by numer- 
ous pits, the larger of which contained lakelets. The 
sides of these pits were steep and exhibited the stratifi- 
cation of the gravel in section, the pits having evidently 
been formed after the gravel was deposited; and there 
can be no doubt that they originated from the melting of 
buried ice and the consequent sapping of the gravel bed. 
The ice whose melting was thus demonstrated may have 
been part of a continuous sheet or may have constituted 
a series of isolated masses, but in either case each local 
mass must have reached its position as part of the glacier. 
The history seems to be, that the waning glacier was so 
reduced by the wasting of its surface near its end that the 
debris of its moraines, worked over by glacial streams, 
overspread the ice and buried a wide belt. At the time 
of my visit this was gradually and irregularly melting, 
under the influence of underground waters, and by its 
melting was sapping the gravel plain. 

This process is of interest to students of Pleistocene 
glaciology, because it is evident that with a slight change 
of conditions it might lead to the formation of familiar 
features of the * modified drift.' If the locality lay some- 
what higher above tide, and if the glacier were so situ- 
ated that, with the progress of its melting, the water of 
ablation would be drawn off in some other direction, the 
gravel plain would be left intact except for such changes 
as resulted from the melting of the ice beneath it. Where 

•n/j/i -[( 

'jfh nl .huB^ lo i'iVKi^ to Lxoilfilutuuj'ji'. biqtn fjib /<! hMhr' 
nobbiH ffio'ri borl*-.!;-//. b/rn^ ^kv/ lnh^inm -^mriudi 'jth b'ji 

.Tjbiil^o ofh ic) ni;q fflBni^ho oiov/ «3««f5fn ooi odt Jxn. .« ji .lu/ 
hHh// qjoja odi ,oIqffiKX'j b'jfraoi-datn'l u «;//od« ^^w^^n^^ '\«i\<f(^ odT 
•jljno^ 'Jib OfilR 8y/od« ll .noitoaa xii Ji^oqab bv*n;o -jdi ^ndidirlza 
Jlf'v '"" -Y ^li ^o aiuiqliJD^ dJoorria oiil fiohiA^o -jdj "io fx|oIfi ffinhrnoJ 
<i I bfiK ?j; so^xjq 008 .vollxiv ]^ru^fii:d ii io ffJuorn od) bnii 
-jffq offifig 3t\l \o o^nU boonBvbn Bfiol n Hwod« Ti*\\4-^\\ *\'»woA^ odT 
io )iao^ob K b'jvi'jD'ji Joqs otll Jyjonomrnoo b/id ;^nH«'j<; lortA .nonoriKMi 

»m the 

quarter of a ^ 
plain in th 


sides oi i: 

cation '^ Kettle-Holes near Hidden Glacier 

Keftle-holes are ascribed to the melting out of ice masses that had 
been buried by the rapid accumulation of gravel or sand. In the 
cases pictured the burying material was gravel washed from Hidden 
Glacier, and the ice masses were originally part of the glacier. 

The Upper Figure shows a fresh-formed example, the steep walls 
exhibiting the gravel deposit in section. It shows also the gentle 
terminal slope of the glacier, the smooth sculpture of its valley wall, 
and the mouth of a hanging valley. See pages 55 and 118. 

The Lower Figure shows a less advanced stage of the same phe- 
nomenon. After settling had commenced, the spot received a deposit of 
mud, and this mud was cracked as the settling proceeded. 

Photographed by G. K. Gilbert, June, 1899. Negatives nos. 371 
and 372, United States Geological Survey. 

student eistoccn 


of CO) 


H. A. E. VOL. Ill 






that ice was approximately or wholly continuous, its melt- 
ing, being temporarily concentrated here and there by 
underground currents, would let the gravels down un- 
equally and leave them, in the end, in a system of irregu- 
lar heaps identical with kames. Where the surviving ice 
comprised only isolated or scattered masses, the waste of 
these would let down the gravels immediately above them, 
producing steep-sided, crater-like kettle-holes, and creat- 
ing the familiar phenomenon of a pitted plain. 

The upper figure in plate v reproduces a photograph of 
one of these fresh-formed kettle-holes, lying near the visi- 
ble portion of the glacier. Its dimensions were 220 feet 
by 70 feet, with a depth, to the water surface, of eight feet. 
Its wall was divided at one point by an outlet channel 
leading to a neighboring glacial creek. An incipient 
kettle-hole seen about a mile below the glacier is repre- 
sented in the lower figure. This had been overflowed by 
water from one of the glacial creeks, so as to receive a 
layer of mud, but the water had retired before it was 
wholly silted up. It is evident that the meandering of 
streams over the gravel plain would eventually obliterate 
all of the kettle-holes, so that their preservation must de- 
pend upon some permanent diversion of the streams. 

The stream escaping from the north edge of the ice 
front was traced backward to an ice cave among the hil- 
locks constituting the lateral moraine. It was evident 
from an examination of the local topography of the ice 
surface that the debris had great influence on the rate and 
method of wasting, and reciprocally, that the wasting 
modified the distribution of the debris. Where the de- 
bris was more than a few inches thick it retarded melting, 
producing an ice hill; but from these hills the stones slid 
down to the neighboring hollows, producing accumula- 
tions which in turn retarded melting and caused a new 
distribution of hills. Some of the hills were elongate, 


and may have marked the position of original moraine 
belts, and one of these happened to be cut across by a 
glacial stream, so as to be exhibited in section. In the 
photograph reproduced in plate vi the ice, which is here 
rather dark from suffused dirt, is seen to constitute the 
mass of the morainic ridge, being preserved from melt- 
ing by a relatively thin layer of fine drift. 

Russell, who saw and named the glacier in 1891, did not 
visit it, and merely records that it was non-tidal. His map 
makes no claim to precision and can not be used in a com- 
parative way to determine the history of change. Gan- 
nett's map, pi. iv, and the various photographs here repro- 
duced, make a record which will be available for future 
comparison, but inference as to past changes can only be 
based on circumstantial evidence. That the recent history 
of the glacier has been one of recession can hardly be 
doubted. Not only do the kettle-holes testify to the stag- 
nation and burial of what was formerly its snout, but I 
found remnant ice masses above the gravel plain on both 
sides of the valley. These were protected from rapid 
waste by gravels that were originally parts of lateral 
moraines, but as water was constantly flowing from them 
their survival could not be indefinitely prolonged, and 
their origin can not have been remote. They were not 
seen more than a half mile from the ice front, but they 
lay considerably above the neighboring gravel plain, the 
extreme height at the north being estimated at 300 feet, 
and at the south somewhat greater. When the back of 
the glacier reached to these heights its front probably 
extended a mile farther down the valley. 

The walls of the valley are not clothed with vegetation,, 
but a scattering growth of annual plants and a few dwarf 
willows have found foot-hold. The gravel plain through 
which the glacial streams meander, though seemingly 
affording conditions of soil and moisture congenial to 

V :n/ 

■ill;/ ^jM ( !. V il!;\ 

.7/ Of I fliifh TJ^irJ 

riiij Ji^jcffit) .X .O vd barlqjn^oJofR 


ffi fi:ioi/;f rn-j/ftiofi 'jlU ^o .">^bh ofli "io ano ^o oiul^inU ..., ,,<.iI8 

)? o^ftq 398 .TjbijIO n^bbiH k) o(iir/f 
! bofJqjn^oJofM 



Upper Figure. — Profile of Hidden Glacier 

Shows the relation of the glacier snout in June, 1899, to topo- 
graphic details of the south wall of the valley. In the foreground are 
bedded gravels that have been overridden, eroded, and grooved by the 
enlarged glacier, and are now being trenched by running \vater. 

The amphitheater high in the mountain shows familiar forms of 
aqueous sculpture but is not continued downward in a gorge of com- 
mensurate size. The gorge probably once existed, but the containing 
spurs were removed by ice erosion at a time when the glacier was much 
larger than now. 

See page 53, and compare figure 28. 

Photographed by G. K. Gilbert, June 20, 1899. Negative no. 36S, 
United States Geological Survey. 

Lower Figure. — Section of Moraine 

Shows the structure of one of the ridges of the northern lateral mo- 
raine of Hidden Glacier. See page 56. 

Photographed by G. K. Gilbert. Negative no. 370, United States 
Geological Survey. 

H. A. E. VOL. Ill 



-»;« --Vy «. 





arctic willows, was absolutely barren (fig. 29), and a sim- 
ilar barrenness was observed on other fresh-formed streams 
of glacial waste. The perpetual stream of cold air flowing 
down from the ice above may be in part responsible for 


The rock fragments washed from the glacier are built into a delta plain, to which addition 
is being made. In the foreground is an alluvial fan encroaching on the plain ; in the dis- 
tance, Russell Fiord. 

this, but I conceive the chief cause to be rapidity of dep- 
osition. The high grade of the plain, in comparison with 
the breadth of its channels and the moderate coarseness of 
its gravel, gave the impression that it was being built up 
with great rapidity. 

As in Russell Fiord and its other appendages, the valley 
walls exhibit the flowing contours of glacial sculpture in 
magnificent development. There are many places where 
the smooth curves of the mountain side, carved out in 
harmony with the flow-lines of the ice, suggest the sweep- 
ing contours of a gigantic ship, rather than the billowy 
backs of a flock of sheep described by the word mou- 
tonnee. A system of flutings can often be traced in sim- 
ple curvature for a half mile, and no one familiar with the 


topography resulting from stream erosion can fail to be 
impressed with the profound modification here wrought 
by the ice. (See fig. 102.) 

As the extent of the glaciers has varied, masses of 
gravel and other drift have been lodged here and there on 
the valley walls and afterward overridden, and where the 
subsequent action has not sufficed for their removal they 
have been carved into forms harmonious and continuous 
with the contiguous rock forms. Here and there, where 
the rills of the valley wall have trenched these deposits 
so as to expose them in section, one may see horizontal 
bedding in a mass of gravel whose external surface ex- 
hibits only the smooth curves of flowing ice. One of the 
larger of these gravel masses lay close to Hidden Glacier, 
against the lower slope of the north wall of the valley 
(pi. v). Upon its sculptured back were scattered boul- 
ders left by the ice which had recently overridden it, 
and among them were a few great blocks of white granite, 
brought from some distant source. Descending toward 
the glacier, the surface of this gravel mass ran under one 
of the remnants of unmelted ice to which reference has 
already been made. 

Nunatah Glacier, — Nunatak Fiord, like the fiord of 
Hidden Glacier, has been boldly sculptured by ice. Its 
lofty south wall descends steeply to the water and is com- 
paratively simple in contour. The north wall is in gen- 
eral lower, is flanked by heavy masses of gravel and other 
drift, and is interrupted by two branching troughs leading 
over saddles of moderate height to the northern part of 
Russell Fiord. These troughs seem to have been largely 
shaped by the ice, which flowed through them to the 
northwest. One is now bare, but the other contains an 
ice mass with the habit of the dying glaciers of Glacier 
Bay. The mass receives a small tributary glacier near its 
summit, but the end seen from Nunatak Fiord in 1899 


Map of Nunatak Glacier 

Nunatak Glacier reaches the sea at the head of Nunatak Fiord, an 
arm joining Russell Fiord at right angles. See map of Yakutat Bay, 
figure 26. 

This map is based on a plane-table sketch by Henr}' Gannett, made 
from stations on the north shore of the fiord, June 20, 1899. Special 
pains were taken to make record of the ice cliff terminating the north 
arm of the Nunatak Olacier. Its south arm, and the minor glaciers, 
were added from photographic data and eye observations, without the 
aid of instruments. 

Drawn by Gilbert Thompson. 

See pages 58-63. 


Y 3nax 







.2 o 
ft 5 

S ft 




bfl 9 
















U O 


08 ^3 


<" rt 

•oi & 


0^ ^ 


^ 52 t>o 

ca (u 


■*-• Vh 'S 


S rt bo 

fl « n 

S 'O M 

O " .« 

T3 ^ O 

fl -O J3 

^'S 3 

,, .-< O 

: S 

2 §5 

2 " ^ 

on a 


S CO 3 

^ ■»-• (0 

rver s 

m ■" -^ 

^ S ^ 

^ * «« 

« w O 

^ '-3 ■« 


^ w 

H S 


was probably stationary and wasting. It was heavily 
coated with drift, which lay in irregular hummocks. The 
opposite end was not identified, but may be one of the 
branches of the stagnant portion of Hubbard Glacier, to 
be described on another page. 

Close to the end of Nunatak Glacier were two lateral 
glaciers which may recently have been its tributaries. 


Beyond it, a hanging valley, with a small glacier. The heights are hidden by cloud. 
Photographed from the southwest, June 21, 1899. 

That at the north (fig. 31) occupied a trough trending 
nearly east and west and intersecting the Nunatak at an 
acute angle. It terminated several hundred feet above 
the Nunatak, its lower part being buried under a heavy 
moraine. That at the south (fig. 32) probably occupied 
a yet higher valley nearly at right angles to the Nunatak 
trough, but clouds cut off the view of its upper portion. 
It was seen only as a series of ice cascades, pouring from 
ledge to ledge for a thousand feet down the steep wall of 
the trough. 

The main division of Nunatak Glacier was tidal, dis- 
charging bergs freely from a cliff nearly a mile long and 
about 200 feet high. At the south the cliff ended against 



a bold rock shore, 
but at the north it 
terminated among 
gravels, and a nar- 
row remnant of 
ice, half buried, [^llfSi 
extended several 
hundred yards 
west of the main 
mass. Near and 
east of the position 
of the glacier front 
is a high rock ter- 
race against the 
south wall of the 
valley, and this 
culminates at the 
east in a bold 
knob, thoroughly 
smoothed and 


rOUnQeO. uy recent The visible portion of the glacier has no valley, but descends 

1 . ,. T fVi wallof valley of Nunatak Glacier. Photographed June 21, 1899. 

hollow separating this knob from the south wall lay a 
mass of ice of uncertain relations (fig. ^^), It was seen 

only from the 
west, and was 
supposed to 
be a tongue or 
arm of Nun- 
atak Glacier. 
The fact that 
it lay several 
hundred feet 




the tidal ariri has raised doubts as to the correctness of 
the first impression, and I now suspect that it was only 
the remnant of a former arm of the glacier, stranded as a 
motionless and slowly wasting summit mass. On the map 
of the Canadian Boundary Commission (1895) it is repre- 
sented as a distributary of the glacier. 

At the time of Russell's visit in 1891 the glacier flowed 
on both sides of the high rock knob (fig. 34) and was re- 
united beyond it, so as 
' '#^^., v--."^;^.--? to convert the knob into 
a nunatak; and it was 
this conspicuous nuna- 
tak near the end of the 
glacier which suggested 
its name. The retreat 
of the ice front in the 
intervening eight years 
can not have amounted 
to less than a mile and 
may have been twice as 
great. It was nearly all accomplished in the first half of 
the period, for the photographs made by the Boundary 
Commission in 1895^ show a complete separation of the 
two arms and a close approximation to the condition of 
1899. The tidal arm was perhaps a third of a mile more 
advanced in 1895, but the non-tidal was not appreciably 
longer. It is possible, however, that the latter extended 
for some distance in a stagnant condition beneath a mantle 
of drift, for at the time of our visit there appeared to be 
remnants of ice in a moraine belt stretching for a mile be- 
yond its extremity. 

The accompanying map (pi. vii) is based on the sur- 
vey by Gannett in 1899, supplemented by photographs. 
It is accurate as to the ice front and contiguous land, but 

^Nos. 20 and 45, on pages 7 and 17 of vol. 17 of the official album. 


Photogfraphed from the north, by D. G. Inverarity 
June 21, 1899. The camera stood on the glacier. 


only approximate in its reproduction of the other portions 
of Nunatak Glacier and the neighboring ice bodies. 

Hubbard Glacier, — The Hubbard Glacier, discovered 
by Russell in 1890 and named by him in honor of Gardiner 
G. Hubbard, president of the National Geographic So- 
ciety, is the most important ice body of Yakutat Bay. 
Its width where it reached the bay was, in 1899, ^^^ ^^^ 
one-half miles, its whole frontage, counting the sinuosities 
of outline, being about six miles. Of this frontage the 
southeastern third was motionless, and fringed for the 
most part by a belt of morainic debris. The remainder, 
pushing itself forward into the head of Disenchantment 
Bay, maintained an imposing ice cliff nearly 300 feet high. 
The active portion of the glacier had two main branches, 
the larger coming from the east or northeast, the smaller 
coming from the north, and the two uniting only three 
miles back from the water. Looking up the valley, we 
could see a number of minor tributaries descending from 
the bordering heights, but the principal sources were con- 
cealed from view, and the low grade of the main trunks 
suggested that their beginnings were far away. The sur- 
faces of both branches were rugged, being divided by a 
labyrinth of crevasses into a wilderness of pinnacles. 
Morainic bands marked out the lines of flow, and a broad 
belt of ice near each margin of the active portion was 
black with included debris. The more southerly of these 
belts was continued to the water front, causing a black 
ice cliff nearly a mile in extent (fig. 35). The correspond- 
ing belt at the north appeared to have become nearly sta- 
tionary, as though resting on a rock shoal, and the flow- 
lines of the northern arm were curved about it. 

The southeastern third of the glacier was moraine 
covered, not only at the water edge but for nearly or quite 
two miles inland. The material was coarse and angular, 
and was divided into zones or belts distinguished at a 



S .S3 


(U S 

CO ;> 

■«-• tn 

<U V 

Q p. 



it! H 

•11 01 


distance by their contrasted colors — black, yellow, purple, 
green, blue-black, and orange or rusty. These bands had 
not the ordinary arrangement of parallel medial moraines, 
but tended rather to contour the slope, and the search for 
their origin and meaning would make an interesting and 
profitable study. Some of them occupied ridges and others 
hollows, suggesting inequality in their ability to retard the 
melting of the ice beneath, but the whole surface was 
rugged in detail, exhibiting a continuous series of hum- 
mocks and kettles. 

The map given in plate viii was made by Gannett from 
a series of stations at the south, the northern shore of Dis- 
enchantment Bay being inaccessible by reason of floating 
ice; and some of the details, especially of the moraines, 
were added by the aid of a series of photographs made 
from a high station on the mainland near Osier Island. 
Russell in 1891 (September 5) stationed his camera on 
Osier Island; and a comparison of his photographs with 
mine (June 22, 1899) shows that there was little change 
in the general character and condition of the ice cliff. My 
views place it one or two hundred feet farther back than 
his, and this difference would probably be increased if the 
proper correction for season could be applied (p. 22), but 
in any case the estimate of the whole retreat would not ex- 
ceed a few hundred feet, and would be very small as com- 
pared to the retreat of Nunatak Glacier in the same period. 

A photograph taken from Haenke Island in 1899 is also 
comparable with one from the same station by the Cana- 
dian Boundary Commission in 1895,^ each showing the 
relation of the northern part of the ice front to the features 
of the contiguous mountain face ; and in this case also the 
recorded change is small. 

Another series of photographs were made by the U. S. 
Fish Commission in 1901, chiefly from Osier and Haenke 

>No. 13, on page 5 of vol. 17, official album. 


Islands ; these show a continuance of retreat. At a point 
where a prominent moraine makes the comparison some- 
what definite, the ice cliff appears to have then stood 700 
to 1,000 feet farther back than in 1899. The cliff was 
also shortened at each end by the enlargement of the 
marginal belts of stagnant ice. 

Turner Glacier, — Turner Glacier comes into Disen- 
chantment Bay from the northwest, a few miles below 
the foot of Hubbard Glacier. As already stated, it was 
tributary to the Hubbard a century ago, and was ren- 
dered independent by the retreat of the latter. Immedi- 
ately after its isolation its front may have projected some- 
what farther into the bay than in later years, but it is not 
probable that the difference was great. A comparison of 
Russell's photograph made from Haenke Island in 1891 
with my own made eight years later from the same station 
(pi. x) shows no appreciable change in the position of 
the front. 

The general width of the glacier within the mountain is 
about one mile, but it begins to flare before fully emerging, 
and at the water front was nearly two and one-half miles 
broad in 1899. For a width of about two miles an ice cliff 
was maintained by the falling of bergs, and the cliff was 
flanked on either side by a sloping tongue which, from our 
distant view, seemed black. Russell's picture represents 
parts of these tongues as white, so that in these marginal 
portions a progressive change is recorded. There was 
also a change in the flow-lines, as indicated by moraines 
near the southwestern margin. A strand of the ice which 
had previously swung far to the south had in 1899 ac- 
quired a more direct course to the bay, reaching the cliff 
1,200 feet to the northward. This would seem to indicate 
that a body of ice near the south end of the water front 
had in the interval become stagnant and, acting as an ob- 
struction, had deflected the current. 

>1q)'f*. .1 

f .h-;.df; 


Upper Figure. — Turner Glacier in 1891 

Photographed by I. C. Russell, September 5, 1891. Negative no. 
556, L^nited States Geological Survey. 

Lower Figure. — Turner Glacier in 1899 

Photographed by G. K. Gilbert, June 22, 1899. Negative no. 317,; 
United States Geological Survey. 

Each of these photographs was made from near the summit of 
Haenke Island, the stations being practically identical. The changes 
in the glacier from 1891 to 1899 are discussed on pages 66-67. 

The position of the glacier, on the west side of Disenchantment 
Bay, is shown in plate viii and figure 26. 


Since writing the preceding paragraph I have been able 
to extend the comparison by examining photographs by 
the Canadian Boundary Commission and the U. S. Fish 
Commission. That by the Boundary Commission was 
taken from Haenke Island^ in 1895, its record being 
midway between Russell's and mine. It shows the con- 
dition of each feature as intermediate between the phases 
of 1 89 1 and 1899. That by the Fish Commission was 
taken from Osier Island in 190 1. It shows an extension 
of the frontal cliff, as compared with the condition in 
1899, and probably indicates renewed movement in mar- 
ginal ice which had become stagnant. 

There is an important morainic belt on each margin of 
the glacier, with outlying ribbons, and a single well-defined 
medial moraine reaches the water front near its middle. 

The visible portion of the glacier within its mountain 
valley has a moderate grade, but at its debouchure into 
the main trough of Disenchantment Bay there is a steep 
descent, the surface falling 500 to 600 feet in a quarter of 
a mile. The grade then suddenly diminishes to almost 
nil, and the glacier terminates in a platform of nearly uni- 
form height, with a width ranging in different parts (1899) 
from 1,800 to 3,500 feet. 

The ice cascade at the point of debouchure indicates a 
drop in the rock bed where the Turner trough joins the 
greater trough of Disenchantment Bay, and this feature is 
related to the phenomena of hanging valleys, to which 
special attention will be given in another chapter. The 
flatness of the terminal portion of the glacier is a peculiar 
feature, not so strikingly exhibited to us in any other in- 
stance. It is of course possible that the longitudinal pro- 
file of the glacier bed is here horizontal, and that the ice is 
everywhere supported by a floor of rock or drift; but it 
seems to me more probable that the flatness is due to 

* No. 13, on page 5 of vol. 17, official album. 



flotation. If the glacier rested on a solid support there 
would be retardation from the friction on its bed, and this 
resistance would tend, under the laws of glacier motion, 
to produce a surface gradient. 

If it be true that the ice is floated, then the sea water 
has access to its under surface, and the rate of melting is 
greater than would obtain if the front only were exposed. 
Should an increase take place in the supply of ice from 
the neve, and consequently in the size and speed of the 
ice stream, the end of the glacier would be thrust farther 
out on the water of the bay, but this extension would in- 
crease the surface exposed to melting, and the loss thus 
occasioned would soon check the enlargement. The op- 
posite result would follow a diminution in the supply of 
ice, and the equilibrium between supply and waste would 
thus be maintained without great modification of the form 
and extent of the ice front. There would of course be 
progressive modification from the silting up of the bay. 
Direct melting of the under surface of the glacier before 
the mass is broken up into bergs would tend to localize 
the deposit of drift, so that the accumulation under the ice 
would be quite rapid, and eventually the drift floor would 
rrr^^^^irTrT^. reach up to the ice 

and subglacial melt- 
ing would be 

If the glacier 
floats, its thickness 
can be estimated 
from the 


FIG. 36. 


from the measure- 
ment of the visible portion. In water of such density 
as Reid observed in Glacier Bay, the ice of glaciers 
floats with about seven-eighths of its mass submerged; 
and the thickness of the visible portion of the tabular 
mass in question would be one-eighth of the total thick- 


ness. The central part of the Turner ice cliff in 1899 was 
250 feet high, but as the surface of the glacier was greatly 
dissected by crevasses its average height above water was 
somewhat less and may be roughly estimated at 220 feet. 
This would give for the total thickness 1,760 feet, and for 
the submerged portion 1,540 feet. The theory that the 
glacier floats, thus implies that the bay has a depth close 
to the northwest shore of 1,600 feet, and the central depth 
should be considerably greater. The theory could there- 
fore be tested by sounding. 

Osier Island. — The little island at the turn from Dis- 
enchantment Bay to Russell Fiord (see pi. viii) is a low 
knoll constituted of the altered shales of the Yakutat for- 
mation. A rocky reef extends northwest from it, and a 
gravel spit, bare at low tide, joins it to the mainland. It 
has three faces, characterized by cliffs telling of active 
erosion by waves. The east face is turned toward Rus- 
sell Fiord and receives the waves generated by southerly 
and southeasterly winds in a straight stretch of deep 
water nineteen miles long and from one and a half to two 
miles broad. A high cliff testifies to their efficiency, and 
so does the gravel spit just mentioned, to which they have 
brought not only pebbles but large boulders. The north 
face, which has a rock cliff of equal or greater height, is 
turned toward the ice cliff of Hubbard Glacier, 6,000 feet 
distant. The wind waves that reach it from Russell Fiord 
have only two miles at most in which to develop. Wind 
waves from the head of Disenchantment Bay, four miles 
distant, might reach it, and on rare occasions they probably 
do, but that part of the bay, being overlooked by the most 
active part of Hubbard Glacier, is ordinarily full of floating 
ice, which prevents the generation of such waves. Instead 
of wind waves the chief attack is by ice-fall waves. From 
four miles of ice cliff the bergs are breaking, and the cannon- 
like boom recording the sundering of one of the greater 


blocks came to our ears every five or ten minutes. As each 
block fell, it started a series of circling waves, many of 
which were so large as to make breakers miles away, despite 
the damping effect of the floating ice. The breakers we 
observed on Osier Island were formidable enough to en- 
force much caution in landing, and the series from different 
ice falls followed one another so closely that there were 
few intervals of quiet. 

The southwest side of the island is sheltered from all 
winds except westerly and, as it borders a cove which 
westerly winds would pack with ice, may never feel the 
force of wind waves. Ice-fall waves reach it from Turner 
Glacier after a journey of four miles, and by shorter, 
but deflected, courses from the Hubbard. Its shore 
cliffs are much lower than those of the other sides of 
the island. 

In 1794, when Hubbard Glacier reached to Haenke 
Island, Osier Island must have been ice-covered and sub- 
ject to glaciation, and it was not bared until more than 
half the subsequent wasting had been accomplished. It 
is therefore probable that the existing shore cliffs, esti- 
mated from memory as 25 to 30 feet high, have been 
carved out within a few decades. During part of 
this time the ice-fall waves reaching the north shore 
were more effective than now because the ice cliff 
was nearer. 

From these various features, and especially from the 
comparison of the north and east shores of the island, it 
appears that ice-fall waves have very notable ability to 
erode coasts, an ability fairly comparable with that of 
wind waves. This fact is of value to the student of Pleis- 
tocene glacial lakes, as it enables him to understand the 
clear outlining of their coasts in cases where the indicated 
extent of the water surface is too small for the generation 
of important wind waves. 


1/ :iiTAJ*^i 

I hnuo8 rriiiil 

inn 'irh Tjnofn b'jtnrf 


:ijii -jiij 1 1 

I ,8t oi ?i 'jy/jviuri; 




Map of Columbia Glacier 

Columbia Glacier reaches the sea at the head of Columbia Bay, one 
of the numerous arms of Prince William Sound (see plate xiii). 

Surveyed by G. K. Gilbert, June 25 to 28, 1899, with use of the 
plane-table, and with stations about the lower end of the glacier. De- 
tails about the nunatak are from notes by Charles Palache. 

Drawn by Gilbert Thompson. 

Contour interval, 250 feet. The bay is represented as at high tide. 
At lowest tide Heather Island is joined to islands north of it by mud 
flats, and extensive shoals are bared along the northeast coast. 

The glacier is described on pages 71-81. It is pictured in plates 
IX and XII, and in figure 37. 


ind wa 



Plate XI 




t f 



Between Yakutat Bay and Prince William Sound we 
made no landing, and our course lay too far from shore 
for observations of value on the glaciers. 

Prince William Sound is an extensive and intricate 
body of water, penetrating a mountain district. Its numer- 
ous islands and peninsulas are mountain peaks or ranges, 
and many of its inner arms and passages have the char- 
acter of glacial troughs or fiords. Among the mountains 
of the mainland at the east and west are many small gla- 
ciers, and a great mountain mass at the north supports 
extensive neves from which magnificent ice rivers flow to 
its northern arms. It was my good fortune to be landed at 
the mouth of one of these ice rivers and, in company 
with Palache, Coville, and Curtis, to spend several days in 
its study. Many photographs were made and some map- 
ping was done. 

In June, 1794, this glacier was seen from the mouth of 
the associated bay by Whidby, one of Vancouver's officers. 
Vancouver says: "To the eastward of this is another 
bay of rather larger dimensions, with an island in its 
northeast corner, . . . terminated by a solid body of com- 
pact elevated ice, similar to that which has been before 
described . . . ; as they passed the eastern bay they 
again heard the thunder-like noise, and found that it had 
been produced by the falling of the large pieces of ice 
that appeared to have been very recently separated from 
the mass extending in vast abundance across the passage 
. . . , insomuch that it was with great diflSculty the boats 
could effect a passage." ^ 

The bay and island appear on a map in Vancouver's 
atlas (see fig. 42). The bay (without the island) is rep- 

1 A Voyage of Discovery to the Pacific Ocean and round the world, etc., Capt 
George Vancouver, vol. v, London, pp. 316-317, 1801. 



resented on a map prepared in connection with expedi- 
tions to Alaska under Glenn and Abercrombie, in 1898,^ 
and the same map also indicates the presence of the glacier, 
but neither bay nor glacier is delineated with sufficient 
accuracy to serve as a record for future comparison. The 
name Columbia was given by the Harriman Expedition. 
The general course of the glacier is southward, and its 
width in the lower ten miles is from three and a half to four 
miles. Its sources are distinct and were not seen, but 
beyond the tract covered by our map (pi. xi) it appeared 
to spread somewhat broadly, and the ice field affording 
its chief supply may send streams in other directions also. 



Shows the western division of the front. 

About nine miles from the sea it encounters an outlying 
mountain over 3,000 feet high, by which it is divided, the 
principal current passing to the west. The eastern arm 
descends steeply for three miles and terminates in a 
land-locked valley against a plain of glacial gravel. A 
subdivision of the western arm enters the same valley. In 
1899 it barely touched the eastern arm, so that the moun- 
tain was wholly surrounded by ice and could properly be 
called a nunatak. Beyond this point the main stream 
flowed to the ocean, but the surface grades descended 
also toward lateral valleys, and there was waste all about 

^Maps and descriptions of Routes of Exploration in Alaska in 1898. U. S. 
Geological Survey, 1899. Map No. 8. 



the periphery. On the east side the ice rested against 
three low hills, beyond which were lakes supplied in part 
from its melting. In one place a lobe of ice touched one 
of the lakes. In a hollow on the side of one of the hills 
a lakelet was imprisoned by the ice, one of its shores 
being constituted wholly by the glacier. On the west is 
an embayment among high mountains, into which the ice 
sent a tongue two miles long, but there was no lake and 
no visible outlet for the water, which must have found 
its way to the sea beneath the body of the glacier. This 
feature is specially remarkable from the fact that the sub- 
glacial water could not follow the course of the ice, but for 



From photographs by W. H. Averell, June, 1899. 

several miles must either move in the opposite direction 
or take some independent route. 

Columbia Bay, to which the glacier flows, is from four 
to five miles broad and is locally divided by a group of 
islands. The western arm, two and a half miles broad, is 
comparatively simple in outline and is probably deep. It 
received the principal discharge from the glacier, which 
spanned it from side to side with a cliff about 300 feet 
high. The eastern arm, irregular in outline and judged 
from the configuration of its shores to be comparatively 
shallow, was bordered by the glacier for a mile and 
a quarter, but the ice cliff was less lofty, and a compari- 
son of its outline with other portions of the glacier showed 


that the water controlled the form of the ice front to only 
a moderate extent. The principal island of the group, to 
which we gave the name Heather, is nearly three miles 
long and has an irregular surface, rising at one point to a 
height of 500 feet. It consists chiefly of rock. Several 
lower islands lie north of it, and the ice front rested 
against one of these for a space of 3,000 feet. The number 
of islands varies with the state of the tide, and it is pos- 
sible that all are united at lowest tide. 

Opposite the great nunatak were two medial moraines, 
one passing within a half mile of its base, the other lying 
about one mile from the opposite edge of the glacier. A 
central tract two miles broad was practically drift-free. 
Toward the end of the glacier this central tract broadened, 
the medials swinging toward the sides, until finally the 
white belt was three miles wide. As the medials diverged 
they also broadened, and they eventually merged with 
flanking moraines, so that near the end, especially on the 
east side, the areas of drift-covered ice were very wide. 
The marginal belt on the west, instead of continuing north- 
ward parallel to the medial with which it was associated, 
was seen to curve about into the western embayment, as 
indicated on the map, and a belt seen from a distance near 
the north edge of the embayment was supposed to be its 
continuation, a loop being made within the embayment. 
As the ice in the embayment descended toward the west, 
it is evident that the morainic loop could not at that time 
represent a line of continuous flow, for we can not suppose 
the ice to flow into the embayment on a descending course 
and then return on a parallel ascending course. It is 
therefore probable that the moraine was formed as a com- 
paratively direct line of drift, following the course of 
the main ice current at a time when no current entered 
the embayment. The inference that a change has oc- 
curred naturally leads to enquiry as to the precise nature 




.iiai morai 


Columbia Glacier, from Heather Island 

The glacier is seen from the summit of the island. Its nearest point 
rests against a smaller island, joined to Heather Island at low tide. 
The part against the land, though steep, is not a cliff. An ice cliff 
iaces the arm of Columbia Bay seen at the extreme right. At the left 
of the small island begins the long ice cliff shown in figure 37. At 
the extreme right, outlined against distant snow, is the hill from which 
the panorama in plate ix was photographed. See pages 71-81 and 
plate XI. 

From a photograph made by E. S. Curtis, June 25, 1899. Nega- 
tive NO. 302. 


of the antecedent condition of the glacier. On the one 
hand, the embayment may have been so full of ice that the 
surface gradient was outward; or, on the other, the glacier 
of the main valley may have had so low a surface that 
there was no tendency to overflow to the comparatively 
shallow side valley. 

The first case implies snow accumulation in the embay- 
ment a few decades ago at a rate not since maintained, 
and would correspond to a general expansion of glaciers 
followed in later decades by contraction; but the rela- 
tions of the ice to the forest, to be described presently, 
show that such contraction has not taken place. The 
second case implies a general expansion of the glacier as 
the important element of its later history. 

Another medial moraine of the great ice field north 
of the nunatak passed just east of the nunatak and 
continued down the eastern arm of the glacier to its 
end, where it contributed toward the building of a 
great alluvial delta which was gradually obliterating one 
of the lakes. 

At the western margin of its principal tidal cliff the 
glacier rested on a bank of drift at the level of low tide, 
and this bank extended eastward as a shoal, on which 
bergs were stranded, for several hundred yards from the 
shore. A bank also extended eastward from the island 
against which the ice front rested, constituting at low 
water a stony cape half a mile long near the foot of the 
ice cliff. These banks testify to a lingering, or linger- 
ings, of the ice front near the position of its modern maxi- 
mum, but it is not easy to estimate the duration of the 
lingering. The western bank is built in deep water, 
but may have been constructed rapidly, as the contig- 
uous portion of the glacier is heavily charged with drift. 
The eastern bank margins a part of the glacier front 
carr}'ing little debris, but occupies an arm of the bay which 



was probably originally shoal. They may have been 
formed recently, or at some earlier epoch. 

The drainage of the ice included several streams 
flowing eastward to the chain of lakes, and we noted 
two important streams from the western ice cliff. One 
of them issued from a cave at the water's edge near 
the western limit of the cliff, the other from a sub- 
merged and invisible tunnel near the middle of the 
cliff. The last mentioned was probably the largest of 
all the draining streams. It rose to the surface at the 
base of the ice cliff and flowed southward over the 

salt water, forming a 
broad lane of milky fresh 
water with a visible cur- 
rent and at times nearly 
free from floating ice. 
At most points the 
forest of spruce and hem- 
lock approached close 
to the ice, its relation 
being similar to that ob- 
served at La Perouse 
Glacier. At the western 
margin of the main ice 
cliff, where the glacier 
crowded against a steep 
rock slope, there was a belt of bare rock, from 200 
to 300 feet broad, between the ice and the forest 
(fig. 38). This belt was strewn with fragments, not 
only of rock but also of wood, and trees were freshly 
overthrown at the margin of the forest. At the time 
of its attack on the forest the ice must have been 100 
feet deeper than in the summer of 1899, and it also 
extended farther southward, as shown by a push-mo- 
raine of rock at the water margin, 800 feet from the 


Shows barren zone and forest. From a 
photograph taken in June, 1899. 



ice front (fig. 39). A second push-moraine, less massive 
than the first, lay within it, being 160 feet from it at the 
water margin 
and elsewhere 
nearer to it 
than to the ice. 
On the island 
between the 
two ice cliffs 
there were 
also two push- 
moraines of 
recent date, 

the nearer being about 100 feet from the ice front, the 
farther from 300 to 500 feet. The latter was associated 
with overthrown forest trees, and included with its rocky 


Photographed in June, 1899. 

debris not only tree trunks and branches but folds of 
peaty soil. The tract between the nearer push-mo- 
raine and the ice was in places occupied by an old mo- 


raine surface over which the ice had advanced, and this 
surface was elaborately fluted in the direction of ice 
motion, the corrugations having a vertical magnitude 
of several feet (fig. 40). In one instance it was seen 
that a large bowlder in the underlying drift had im- 
pressed its form on the ice, preserving in its lee a train 
of drift of the same cross-section, which constituted 
a ridge, and it is probable that the other flutings were 
of the same character. As these details in the con- 
figuration of the drift surface would be quickly ob- 
literated by frost and rain, their exposure must have 
been very recent. Probably the advance creating the 
push-moraine and the subsequent melting which laid 
bare the ice-molded drift had taken place within one 
or two years. 

On the mainland at the east the same phenomena were 
observed, with the exception of the fluted drift surfaces. 
There was an inner push-moraine, chiefly or wholly of 
drift and running parallel to the ice margin. There was 
an outer push-moraine, less regular in its distance and 
associated with disturbance of the forest and the meadow 
peat (fig. 41).^ In the tract between the two many 
prostrate trunks were seen, showing that in places the 
front of the forest had been crowded back several hun- 
dred feet. Many of the trees that were overturned but 
not overridden, retained their bark, branches, and even 
minor twigs, but the leaves had fallen. On disturbed 
forest soil Coville found three young spruces which 
had grown since the catastrophe. In each case the 
age, as shown by rings of growth, was seven years. 
The date of the ice maximum was therefore not later 
than 1892 and may have been that year. 

iThe view in fig. 41 is toward the northeast — along the front of the push- 
moraine. A little of the steep face of the glacier is seen at the left. At right 
is a tract of undisturbed bog. 







The overturned forest trees associated with the push- 
moraines on the eastern and western shores of the bay and 
on the island, exhibited the same general appearance of 
recency, and there can be little doubt that they were dis- 
turbed at the same time. They demonstrate a temporary 
increase in the size of the glacier, not of precisely the 
same amount at all points, but of the same order of mag- 
nitude. Previous to that expansion the glacier had been 
smaller during a period at least sufficient for the growth of 
the overturned trees. The evidence from forests and 
push-moraines does not show whether the ice during this 
epoch stood continuously near to the forest or was sub- 
ject to wide oscillations in extent; but the bending of 
the moraine belt on the back of the glacier into the 
western embayment (page 73) gives strong support to 
the view that the recent maximum was preceded by an 
important minimum. 

No attempt was made to estimate the age of the trees 
by counting rings of growth, but the forest had the char- 
acteristics of maturity, and the time required for its pro- 
duction could hardly have been less than two or three 
centuries. The mountain side just west of the glacier, 
rising steeply to a height of 2,000 feet, is clothed with a 
luxuriant growth from the push-moraine up to about 1,500 
feet. Many of the trunks are three or four feet in diam- 
eter, and among them lie prostrate logs in a state of de- 
cay. Upon the islands, and on the lowland near the east 
margin of the glacier, the trees are somewhat smaller, but 
the many dead trunks standing among them indicate that 
they are mature, and their term of life may be as long as 
that of their western neighbors. 

A further item of information as to variation, albeit 
somewhat indefinite, may be derived from Vancouver's 
map. It is not sufficiently precise to affi^rd identification 
of any topographic detail of the bay except Heather Island 

iiix a¥Aj<i ^Ito ^or^AviAj^za 

••'j yovHir^. ji^iio'J «'jju}8 Iji):)rnU no ^^ftairb h9<*i; 

-lit) (Ci XlWK'iCI .v^)v 

.eriqL-ioo:torfq ;to Or^IiJ obi;m «£7^ ons bciK : .A 


Upper Figure. — Prince William Sound 

An index map of the sound, showing positions of Port Wells and 
Columbia Bay. Based chiefly on United States Coast Survey chart 
NO. 3091. 

Lower Figure. — Sketch of Port Wells 

Surveyed by Henry Gannett, June 26 to 29, 1899. Drawn by Gil- 
bert Thompson. Details about Amherst Glacier are in part from the 
map of an expedition under Captains Glenn and Abercrombie, U. S. 
A. ; and use was made also of photographs. 

ear the 

ay dcaU U ' 



Plate XIII 





(see fig. 42), but the narrowness of the strait represented 
between the island and that portion of the coast said to 

• vs^N^ -^-.^r V- : ^^^sist of ice, in- 
^" '^ • dicates the impres- 

sion of the explorers 
that the ice wall 
stood not very far 
beyond the island; 
and this view is 
supported by the 
statement already 
quoted that the 
island was in the 
" northeast corner " 
of the bay. It seems 
reasonable to infer that the glacier was not much smaller 
in 1794 than in 1899; and that if the features of the em- 
bayment prove a recent and important minimum, that 
minimum occurred in the nineteenth century. 

A, enlarged from Vancouver's map (1794), which does 
not distinguish the glacier from other parts of the land. 
£, reduced to same scale from plate xi, showing relations 
of sea, glacier, and land in 1899. 


While our boat party was occupied with Columbia 
Glacier the main division of the Expedition visited the 
northwestern arm of the sound, called Port Wells, where 
important contributions were made to geographic knowl- 
edge. College Fiord, the right branch of Port Wells, was 
explored more thoroughly than ever before, and the left 
branch, Harriman Fiord, was discovered as well as ex- 
plored. The fiords were mapped by Gannett, and their 
beautiful and imposing series of glaciers were photographed 
by half a dozen cameras. After the ship had picked up 
my party in Columbia Bay, it returned to Harriman Fiord 
for Gannett and Muir, and I was thus enabled to sail past 
several of the Port Wells glaciers, but the following de- 
scription is chiefly at second hand. Many of the best 


photographs of the Port Wells glaciers are reproduced by 
photogravure to illustrate the narrative of the Expedition, 
and, to avoid needless repetition, I have selected for my 
own use only the views most important in connection with 
my text, but I shall refer freely to the plates of volume i. 

College Fiord is from two to three miles broad and 
about twenty miles long, trending north-northeast and 
south-southwest. Near the south end, where it joins the 
main body of Port Wells, there is a bay on the east side 
overlooked by two non-tidal glaciers. The larger of these 
was called Amherst by the Expedition, the name being 
given in honor of an American college. Somewhat north 
of the middle the fiord sends an arm to the northeast, and 
this arm receives a large tidal glacier, the Yale. At the 
head of the main fiord is the Harvard, also a large tidal 
glacier. Several branches of the Harvard were visible 
from the ship, and that next to the ice front on the north- 
west was named Radcliffe. A series of glaciers on the 
northwest side of the fiord resembled the Radcliffe in 
general character, and four of these received names — 
Smith, Bryn Mawr, Vassar, and Wellesley. 

Amherst Glacier was passed by the ship at some dis- 
tance, and its features are known chiefly through the pho- 
tographs secured by Merriam (pi. xiv). It is fed by 
neves in full view from the fiord, and approaches the sea 
in a short, broad stream which at first descends steeply 
and afterwards more gently. The habit of the lowland 
lying between the glacier and the ocean indicates that it 
is built of morainic material. Near the sea is a belt of 
timber, but this is separated from the ice by a barren tract 
similar to that about Davidson Glacier. A barren zone 
several hundred yards broad is seen to flank the glacier on 
the southwest, and a similar zone borders its companion, 
Crescent Glacier. These features doubtless indicate shrink- 
age in modern times, the change having been of moderate 

YI7 :rry.j 


'.^ :-,.'. /n:il .J 



Amherst and Crescent Glaciers 

The Amherst Glacier is at the left, in the view, the Crescent at the 

These glaciers are on the east side of College Fiord, at its junction 
with the main body of Port Wells. Their position is shown in plate 
XIII. See pages 82—83. 

From a photograph made by C. Hart Merriam, June 26, 1899. 
Negative no. 124 of the United States Biological Survey. 

A bar 



amount, although greater than in the case of La Perouse 
and Columbia glaciers. The Crescent is comparatively 
narrow, and approaches the sea with a higher grade. A 
curve in its trough conceals its upper course. 

The Yale drains a larger area and receives a number of 
tributaries. The front of the cliff is wide, but of moderate 
height, and a blackening, west of the middle, by englacial 
drift suggests that a rock knob may lie near the surface, 
ready to develop into a nunatak or island if the glacier 
shall diminish. The trough in which it lies is forested 
along the water edge on both sides for the greater part of 
the distance :^. 
from the main 
fiord to the 
glacier, but 
barren in the 
vicinity of 
the glacier. 
There are 
trees high on 
the valley 
wall at the 
end of the 
glacier, but 

they do not come down close to the ice. An excellent 
" photograph of the glacier is reproduced at page 128 in 
volume L It is enlarged from negative 113, U. S. Biolog- 
ical Survey. A nearer view of the eastern part of the 
front (fig. 43), although lacking detail at the critical 
point, may serve a purpose for comparison when the gla- 
cier shall be revisited. It shows a small tributary, cas- 
cading from a hanging valley near the end of the main 

Shows position of front in 1899 in relation to a tributary from the east. 


The Harvard Glacier is by far the greatest discharging 
to the fiord. Only a few miles of it are visible from the 
sea, as the main trunk, traced backwards, curves to the 
right and disappears. So far as visible it is of low grade, 
its slope being gentler than that of any other we saw ex- 
cept the Muir and Columbia; and this feature, taken in 
conjunction with the notable height of the cliff in which 
it terminates (350 feet), indicates great depth of the ice 
stream and a remote source. The large number of medial 
moraines tells us that it has many branches, and five are 
visible from the sea. One from the southeast joins within 
a mile or two of the end, and the other four come from 
the northwest and north. All have steep grades in ap- 
proaching the main trunk, but two at least show gentler 
grades at higher altitudes. The Radcliffe joins the Har- 
vard so close to the water front that it does not become 
fully merged with the greater stream, but merely coalesces 
at one edge on its way to the sea. A conspicuous medial 
moraine of the Radcliffe maintains its high declivity quite 
to the water's edge, and the cliff where the Radcliffe 
ends is notably lower than the confluent cliff along the 
front of the Harvard. The next tributary is likewise 
characterized by a strong medial moraine, and the branch- 
ing from which this arises can be seen a short distance 
back from its junction with the Harvard. The photo- 
graphs show no trees in close proximity to the Harvard. 
The point of land at the branch in the fiord five or six 
miles to the south is forested, and this forest follows the 
coast for some distance toward the glacier, but stops sev- 
eral miles from its front. The opposite coast is reported 
free from trees at the water's edge for eight or ten miles, 
but at an altitude of several hundred feet the trees ap- 
proach the glacier. These relations, while they do not 
show whether the glacier is now waxing or waning, indi- 
cate that its length has been several miles greater at a 






2 3 

•o o 


•^ vi 



5 -a 

■" bo 


O V 


ft 5!S 
^^ ft 

fl ° a 

ft ft ft 

ft H *^ 
« *i 


•*3 .S ;3 

^ 2^ 


period so recent that the forest has not yet reoccupied the 
abandoned ground. 

As the Harvard has recently retreated more than its 
neighbors, and as its velocity is presumptively high, its 
future changes will have special value as indices of the 
local variations in the conditions of glaciation, and the 
record of its magnitude in 1899 is therefore important. 
Unfortunately the party did not approach it closely, and 
the photographs are deficient in detail at critical points, 
but they will enable the future observer to recognize any 
change of important magnitude. The best available data 
for 1899 are contained in Curtis's photograph No. 273, 
reproduced at page 72 in volume i. The east end of the 
frontal cliff, as there shown, is not far (between 1,000 and 
2,000 feet) from the apex of a delta, or alluvial fan, built 
by the stream from a small hanging glacier. It is also 
seen that the eastern part of the glacier cascades about a 
half mile from the front, dropping so low that its tidal 
cliff has only half the height of that of the central part. 
At the west the relation of the Harvard to the Radcliffe 
is manifestly sensitive to the influence of advance or re- 
treat. A moderate recession would separate the two 
glaciers ; an advance would deflect the lower part of the 
Radcliffe medial moraine into parallelism with the Har- 
vard medials (see fig. 44). 

Smith Glacier reaches the fiord three or four miles from 
the Radcliffe, and is of the same order of magnitude. Fed 
by several tributaries among the crests of the range, it 
gathers in a high mountain valley and then descends in 
magnificent cascades down the mountain front to the sea. 
In the last part of its course it has scarcely any valley, 
the outer surface of the ice being practically flush with 
the face of the mountain ; and there is no flattening of its 
profile as it reaches the water. Though its lower slope is 
so seamed by crevasses as to exhibit a mere congeries of 



to . 


a t 

3 bo 

V 'o 

o — ^ 

o n ^ 

^ * g 

i Ji^ 


pinnacles, two lines of medial moraines are distinctly 
traceable, each partitioning off a fourth part of the ice 
stream at the side. 

The Bryn Mawr, next south of the Smith, is somewhat 
larger. Its two main branches, gathering in mountain val- 
leys not well seen from the sea, become visible in twin 
cascades, and then, uniting their streams, make a second 
leap to the sea. As tide is reached, there is a tendency to 
flatten the profile, and the central portion of the stream 
becomes nearly or quite horizontal for a few hundred feet 
before breaking off in the terminal cliff. 

Next in the series comes the Vassar, parallel to the 
Smith and Bryn Mawr and exhibiting a similar series of 
cascades, but of smaller size and less direct in its course. 
It is cumbered, especially in its lower part, by rock debris, 
and close inspection was necessary to determine the fact 
that it was actually tidal. 

The Wellesley, last of the tidal series, flows with gentle 
grade through a mountain trough joining the fiord at right 
angles, and then cascades to the sea, into which it plunges 
without notable modification of profile. Beyond it are 
small glaciers occupying alcoves on the mountain front 
but ending far above the water. 

The Bryn Mawr, Smith and Radcliffe are represented 
in fig. 45, the Bryn Mawr alone in the frontispiece, and the 
Wellesley in a plate at page 122 of volume i, from a pho- 
tograph by Merriam (No. 121, U. S. Biological Survey 
series). The Bryn Mawr was photographed at shorter 
range by Curtis (No. 276 a), but the view has not been 

In the intervals between the tidal glaciers just described 
there is no forest at the water's edge, and the photographs 
reveal none at higher altitudes ; but a little farther south 
the coast is forested, and the trees climb up a few hundred 
feet on the moraine heaps under the hanging glaciers. 


They are separated from the ice, first by a broad belt of 
alders, and then by a barren zone. As the spruce forest 
in College Fiord nowhere stands close to the ice, but is 
separated by a barren zone, it seems fair to assume that 
the ice has occupied this zone so recently that the period 
since its shrinkage has not sufficed for reforesting; but no 
facts are recorded tending to show the nature of the 
changes immediately preceding our visit. 

The four tidal glaciers on the northwest side of the 
fiord, and four branches of the Harvard reaching it from 
the same mountain range, show a remarkable agreement 
in certain general features of profile. Near their de- 
bouchure they descend in one or more steep cascades 
through a vertical space of 1,700 to 3,000 feet, and back 
of these cascades their slopes are comparatively gentle. 
Their upper valleys are deeply incised, but their lower 
valleys are shallow, barely sufficing to hold the ice 
streams, so that the faces of the glaciers are nearly flush 
with the general face of the fiord wall. These features 
indicate that the principal work of ice sculpture was per- 
formed when the trunk glacier filled the fiord to a level 
somewhat above the line of cascades. It was then that 
the fiord wall received its smooth contours, and much of 
the general excavation of the fiord may have been per- 
formed at the same time. The tributary ice streams from 
the side carved shallower troughs, adjusted to their needs, 
and were prevented from excavating deeply at any point 
because the great trunk glacier gave them a high base- 
level of discharge. These tributary troughs will receive 
further consideration in the chapter on Pleistocene glacia- 


From the bend where it is joined by College Fiord 
Port Wells extends only three or four miles northward 
and is reduced in width. Its trough then swings quickly 



to the west and southwest and is continued twelve miles 
farther. At the apex of the turn a large glacier (the 
Barry) protrudes from the northern shore, reducing the 
waterway to a narrow strait. The portion above the 
strait, having been discovered by the Harriman Expedi- 
tion, was named Harriman Fiord. The general width of 
the fiord is from two to three miles. Considered as a 
mountain trough, it branches near the middle, but the 
western branch is almost wholly occupied by a gla- 
cier. Its walls are everywhere high, and it is in fact 
a secluded pocket among the mountains. All about 
are glaciers, of which four are of large size and six 
reach the sea. 

Barry Glacier (fig. 46), at the entrance to the fiord, 
approaches from the north-northeast. Its low grade indi- 
cates a distant 
source, but 
the source 
was not seen, 
as its upper 
valley was 
concealed by 
mist. Unfor- 
tunately the 
map data se- 
cured do not 


afford an ac- 

The glacier comes from behind the dark hill at the right. The ^,,*-of p Ap^fpr 
visible tributary descends by two cascades. Doran Strait lies be- *^UlctLC UCLCI- 
tween the glacier front and the sloping point of land at the left. mination of 

the dimensions of its end, but it impressed the beholder 
as one of the largest ice rivers of Port Wells. Its peculiar 
relation to the fiord causes it to be swept by the passing 
tide and prevents the accumulation of icebergs about its 
front, but the same relation exposes it to exceptionally 
rapid melting by the sea, and the conflict of ice current 



with tidal current must be active. The forward flow of 
the ice tends to narrow the strait, and this constriction, by 
increasing the speed of the tide, enhances the melting 
power of the water. The fact that the glacier was able 
to occupy two-thirds of the width of the fiord indicates 
that its forward movement was strong. 


Showing the relation of a medial moraine to a great oblique dirt band. Photographed by 
E. H. Harriman from the ship, June, 1899. 

Its moraines were of small relative importance, but 
a belt along the western margin was darkened by drift 
and there 
were two 
medials. One 
of the latter, 
exhibited in 
section in 
the face of 

the cliff, was seen to be the surface outcrop of a sheet 
of drift-charged ice which extended obliquely down- 
ward, passing under the western portion of the stream 
(see fig. 47). 

The cliflf was further diversified by a number of caves 
at the water's edge, supposed to be the mouths of englacial 

FIG. 48. 

From a photograph by W. R. Coe. 



Connected with the eastern edge of the ice was a long, 
narrow tongue attached to the shore (fig. 49), evidently a 
remnant left by the glacier at some very recent date when 
its front was more extensive. As this strip was not pro- 
tected by gravel, it must have been wasting rapidly, and 
the period of its separation may have been only a few 
months. I was in doubt whether to ascribe it to a pro- 
gressive shrinkage of the glacier or to seasonal variation. 

Showing associated remnant of stagnant ice. From a photograph by D. G. Inverarity. 

On the same coast the forest did not approach the glacier 
closely at the water line, but passed above it, leaving a 
barren zone several hundred feet broad. The common 
boundary of the barren zone and forest was so well defined 
as to indicate that it represented a former limit of the ice, 
but there were no overturned trees. If the forest ever 
occupied the barren zone, and was there destroyed by an 
advance of the glacier, the occurrence was so long ago that 
the overturned tree trunks had disappeared through decay. 
The portions of the forest nearest the ice included no 
trees of large size, but as there were many standing dead 
trunks it is probable that the growth was mature and that 
the small size of the trees indicated merely conditions un- 



favorable to luxuriant growth. Translating these facts into 
terms of glacial history, it seems probable that the Barry 
had been, at some time within the century, somewhat 
larger than when we saw it, but that it had not for a series 
of centuries exceeded the limit marked out by the neigh- 
boring forest. If any change had occurred within the 
last year or two it was of diminution. 

The opposite wall of the fiord is forested down to the 
water's edge, and it is thus shown that no recent advance 
of the glacier has carried it completely across the channel. 

Next west of the Barry is Serpentine Glacier, coming 
down to the fiord from the north. It is a broad stream, 
of low grade, fed 
by four or five 
tributaries de- 
scending steeply 
from amphithea- 
ters in the encir- 
cling mountains. 
Though it reaches 
the sea, it yields 
few bergs, but is 
building a mo- 
raine barrier 
along most of its 
front. Its medial 
and lateral moraines are conspicuous, especially the north- 
ern lateral. Like the Turner and Reid, it seems to rest 
on a valley floor considerably above the floor of the 
fiord to which it is tributary. Its most westerly branch 
(fig. 50) heads in a high valley not fully commanded from 
the fiord and falls to the main glacier in two fine cascades. 
At the level of the upper cascade, 3,000 to 4,000 feet above 
tide, are three hanging glaciers, perched in alcoves of the 
valley wall where it curves to join the wall of the fiord. 


The main body of glacier lies behind the nearer hill at right. 



The only observed fact bearing on its recent history 
of change is the absence of trees from the valley walls 
near it. 

Two of the hanging glaciers are shown, at the left, in 
figure 50. The cascading tributary and other branches 
of the Serpentine appear at page 124 of volume i, in a 

plate repro- 
ducing photo- 
graph NO. 292 
of Curtis's 

Glacier reach- 


Surprise Glacier (at right) has its source in a valley system be- Irom the WCSt, 
yond Cataract Glacier. i 

and occupies 
a deep and long branch valley. Its sources were not 
visible, being concealed by the curvature of its valley, 
but its moderate grade and the lofty ice cliff in which it 
ends, mark it as an important ice river. 

Its near neighbor. Cataract Glacier, occupies a narrow 
and lofty mountain trough, from the end of which it sends 
a steep, tapering tongue down to the sea. It is interme- 
diate in type between the hanging glaciers of the Ser- 
pentine valley and the cascading glaciers of the west wall 
of College Fiord. 

The valley containing Harriman Glacier is a continua- 
tion of the main trough of the fiord and holds the same 
general southwest trend. The glacier curves toward the 
west and then toward the south, disappearing from view 
at a distance of nearly ten miles. As the most distant 
portion seen has a gentle slope and lies far below the 
bordering mountains, it is probable that the sources are 
still several miles beyond. Its general width is about a 
mile and a half, but its high-grade tributaries are so thick- 

/ "/ 

. lo 
:)kIO nntuh 

Jfi^^il^ i; fd tibia vslljiy 

bfim rlqiJi^oJodq fi rrroi'i 
.-"itr i-' -iiJiuO '.iih iu b'dL .oz ■j/hn'^oVl 
Ut. bfiii fiil'j biJnoii odl to ijii'j rltion ydl ^noiU »>'^^o^^^ n^'iroA oriT 
fujom sdJ no ri^^iH .vollfiv jrh to lb:// j^-jv/dfion odi k) f'txii^i ■^i\uno\ 
'■{.) ^o;>Rq oori .loiod^ [' ; I ' ' < e^'^yHfiV ^iii^mul ov/l 'iij; riiut 

' ' ' '■<-> iJr8c^jg0^^tK23>I 

M^8 1 

iij{^ f/JhiiTJvrj 

:ic han: 

The cascading t 


Harriman Glacier 

The Upfer Pigu7'e shows the south end of the frontal cliff of the 
glacier and adjoining parts of the valley wall. The main part of the 
glacier lies outside the view at the right (see plate xiii). At the left 
a tributary' glacier descends from the mountain, and there are glimpses 
of other tributaries farther to the right. The visible part of the Har- 
riman Glacier is probably a motionless remnant, left clinging to the 
valley side by a slight retreat of the central part of the frontal cliff. 
See pages 94—96. 

From a photograph made by D. G. Inverarity, June 27, 1899. 
Negative no. 286 of the Curtis series. 

The Lower F'igure shows the north end of the frontal cliff and ad- 
joining parts of the northwest wall of the valley. High on the moun- 
tain are two hanging valleys, each with a small glacier. See pages 95 
and 119. 

From a photograph made by D. G. Inverarity, June 27, 1899. 
Negative no. 285a of the Curtis series. 


. mc 
esare s 


Harriman Glacier 


.4^'>^< • ^^ 


Harriman Gla.cier 



set as practically to coalesce, especially on the south- 
east side, giving a broad expanse of nearly continuous 
ice and neve. This expanse, fully commanded from 
the water, makes the view of the glacier a most im- 
pressive spectacle. 

The visible moraines are few and unimportant, but the 
presence of much embedded drift is suggested by a detrital 
bank on which the eastern edge of the ice is seen to rest 
at the front. Above this bank the frontal cliff is low and 
irregular, but elsewhere it is lofty, ranging in height from 
200 to 300 feet. From such a cliff an active discharge 
of bergs might be assumed, but our parties encountered 
only a moderate quantity of floating ice near the head of 
the fiord. 

The glacier is not closely approached by forest growth, 
but shrubs were seen on the shore of the fiord within a 
few hundred yards of the ice. If the ice is diminishing, 
the recent retreat of the glacier front would appear not to 
have been rapid. The condition of the front in June, 
1899, is recorded in a series of photographs. Two of 
these, reproduced in plate xv, show the ends of the frontal 
cliff, where the ice adjoins the valley walls, and will be 
serviceable for future comparisons with reference to ad- 
vance or retreat. A third (Curtis negative no. 291), 
published in volume i at page 74, gives a distant view of 
the glacier and its southeastern tributaries ; and a fourth 
(Harriman negative no. 98), appearing in volume 11 at 
page 262, gives the glacier and its surroundings from a 
somewhat nearer point. 

Some of the minor glaciers associated with the Harri- 
man occupy elevated valleys far above the main trough, 
and these upland valleys probably constitute a system 
initiated at an earlier epoch, when the fiord was flooded 
with ice to a great depth. Illustrative examples are 
afforded by two hanging glaciers overlooking the lower 


part of the Harriman from the northwest (pi. xv). No 
measurements were made, but it is evident from an inspec- 
tion of photographs that the heights of such features in 
this neighborhood are approximately the same as in the 
vicinity of Serpentine and Surprise glaciers, and it is 
possible that a number of minor glaciers observed on both 
sides of the fiord constitute with these a general system. 
Roaring Glacier, between the Cataract and the Harriman, 
owes the peculiarity suggesting its name to an abrupt 
change of grade. From a comparatively gentle slope it 
passes to one so steep that loose masses find no lodgment, 
and as its movement steadily projects its end beyond the 
point of inflection, fragments of ice break away and tumble 
down the steep incline, to gather in a heap far below, where 
they lie until melted. 

The condition of extreme glaciation to which these phe- 
nomena point does not belong to the series of modern 
changes, and will be referred to again in the chapter on 
the Pleistocene history. Were it of comparatively recent 
date the fiord would now be destitute of trees, but such is 
not the fact. It is true that the slopes are bare in the 
immediate vicinity of the glaciers, and that the valley 
walls enclosing the greater glaciers — the Barry, Serpen- 
tine, Surprise and Harriman — support no trees, but the 
lower parts of the fiord walls are elsewhere covered by a 
hemlock forest. 

As to the proper interpretation of the peculiarities of 
forest distribution the case is not altogether clear. In 
other localities there has seemed good reason to ascribe 
absence of forest to recent occupation by ice, but here 
there is a sort of transition from forest to barren which 
suggests climatic limitation. In the zone of transition the 
trees are not young and vigorous, as when invading newly- 
acquired territory, but scrawny and ill-favored, as though 
struggling desperately against the attack of hostile condi- 



tions. In the accompanying illustration, representing 
the edge of the forest nearest to Harriman Glacier, 
the rareness 
of branches 
on the side 
toward the 
water sug- 
gests that 
winter fogs 
driven land- 
ward over- 
whelm the 
boughs with 

loads of ice. fig. 52. hemlocks bordering harriman fiord. 


West of Prince William Sound we saw glaciers in 
abundance — on Kenai and Alaska peninsulas and on 
Unimak and Unalaska islands — but only one was visited 
or approached so closely as to permit the making of ob- 
servations worthy of record. 

The Grewingk Glacier is one of a series descending the 
northwest slope of the mountainous Kenai Peninsula. It 
was mapped by Dall in 1880, and revisited in 1895, when 
he made accurate note of the position of the ice front with 
reference to an object on the southern wall of the valley. 
In 1899 he accompanied me to the edge of the glacier for 
the purpose of pointing out this object, and I made a few 
photographs and other observations to aid in the recog- 
nition of future changes. 

The glacier descends with moderate grade from a high 
neve, and maintains in its lower part a width of one and 
one-third miles. Its front is three miles from the sea, the 
interval being occupied by a gravel plain. For half the 


distance the plain is bounded at the sides by rock walls, 
continuous with those containing the glacier, but these 
end at the general line of coast, and the plain flares beyond 
them as a low, broad cape. It is in fact a delta of glacial 
detritus, filling the lower part of the glacier trough and 
encroaching on the bay. The building of the plain is 
rapid. Its upper part was almost barren, as we saw it, 
only supporting enough scattered young spruces to show 
that their spread was not absolutely prohibited by soil or 
climate. Lower down were plantations of vigorous young 
cottonwoods, but no mature groves were seen. Border- 
ing lands of earlier origin are covered by spruce forest, 
and in places the growing gravel deposit was evidently 
invading the forest, overwhelming the undergrowth and 
burying the roots of large trees so that they languished 
and died. I was impressed with the fact that the quantity 
of rock waste discharged by the glacier was much greater 
than would normally be discharged by a stream of water 
draining a similar area. 

The glacier bore no large moraines, and its generous 
output of rock waste must have been supplied chiefly by 
the englacial drift. The visible belt of this drift was 
broad at one or two points, but in general so narrow as to 
give the impression that the base of the ice la}^ consider- 
ably below the level of the gravel plain. The ice front 
was steep, probably ranging from 20° to 30°, and im- 
pressed Dall as much steeper than in 1895. It was de- 
cidedly steeper than the front of Hidden Glacier and the 
north front of the Hugh Miller, observed a month earlier, 
but less steep than non-tidal portions of the Columbia. 
If the correlation of high and low frontal slopes with ad- 
vance and retreat is well founded, the Columbia and 
Grewingk glaciers were advancing in 1899 and the Hid- 
den and Hugh Miller were retreating. If the slopes are 
related to the direction of the sun, those toward the south 





are steeper than those toward the north. Close to the ice 
front the plain was interrupted by a low ridge of gravel 
(see fig. 54), a push-moraine associated with some small 


and recent advance of the glacier, and possibly a phenom- 
enon of the annual oscillation of the front. Farther out 
on the plain the course of a much larger moraine was 
marked by a crescentic line of mounds and short ridges, 
remnants of a once continuous morainic rampart that had 
been breached at many points by streams from the gla- 
ciers. I judged that this also was a pushed-up ridge, 
resulting from a plowing of the gravel deposit during a 
rapid advance of the glacier. It was not strictly parallel 
to the ice front in 1899, nor to its outline as mapped in 
1880, but was more convex downstream. Its middle 
part was 2,600 feet from the glacier in 1899, and its most 
southerly remnant 800 feet. In 1880 Dall observed a 
considerable amount of ice under the sand and gravel of 
these mounds, and I noted many kettle-holes resulting 
from recent melting of ice remnants under the gravel 


plain inside the crescent. It is therefore probable that the 
advance occurred only a few decades ago. 

Between 1880 and 1895 the ice retreated through a 
space estimated by Dall at 250 feet. Between 1895 and 
1899 the portion of the front adjacent to the south wall of 
the valley retreated 350 feet. It happened that all of the 
observations were made in the same month, so that the re- 
corded changes were independent of seasonal fluctuations. 

Figure 54, representing the south wall of the valley at 
the front of the glacier, shows the position of the ice on 
July 21, 1899. The gravel plain and the inner push-mo- 
raine occupy the foreground. At the extreme right a low 
boss of rock juts from the valley wall into the gravel 
plain, and just to the left of it is a large boulder resting on 
a drift terrace. On July 29, 1895, Dall found barely space 
to pass between the glacier and this boulder. 

The same view shows the barren zone of the valley 
side and the lower limit of the forest, and the correspond- 
ing features for the north side of the glacier are seen 
at the left in the general view of the glacier, figure 53. 
At the south the line of forest was about 200 feet above 
the edge (1899) and descended westward to the gravel 
plain in 800 feet. For some distance below the line scat- 
tering young spruces and other bushes were seen, and 
even within the tract uncovered since 1895 a few alders 
had started. Nevertheless, the line of forest was clearly 
defined, being the lower limit of mature trees, dead trees, 
and accumulated humus. Though the line was unques- 
tionably caused by an advance of the ice, it was not 
marked by heaps of overturned trees, as at the La Pe- 
rouse and Columbia glaciers. 

From these various phenomena a number of inferences 
may be drawn: (i) For several centuries — the age of the 
forest, including its dead trees — the glacier has not ex- 
tended beyond the lower limit of the forest; that is, it 


has not very greatly exceeded its present size. During 
that period it may or may not have been much smaller 
than now. (2) During the latter part of that period the 
glacier advanced to the forest line and then retreated. 
This maximum probably occurred as much as fifty years 
ago — to allow for the disappearance of the overturned 
trees — but could hardly have been so early as the begin- 
ning of the nineteenth century, else there would be larger 
spruces below the forest line. (3) Several decades ago 
there was a maximum, affecting especially the central part 
of the glacier, and retreat from this was still in progress 
to the close of the century. 

There was little difference in the extent of the two 
maxima, and although their separateness was not doubted 
during my visit, it now seems to me possible that the two 
were identical. One was inferred wholly from the fea- 
tures of the valley wall, and the other wholly from the 
frontal moraine. The straggling of young spruces below 
the forest line afforded so strong a contrast to the absolute 
barrenness of the morainic mounds that the possibility of 
connecting the two groups of phenomena with the same 
event did not occur to me; but Dall's observation of ice 
remnants in the moraine in 1880 suggests a local cause 
for the sterility of the gravel mounds and leaves the 
matter in doubt. 


During the last twenty years much attention has been 
given to the variations of glaciers, and a large body of 
facts has been collected, especially with reference to 
European examples. In attempts to generalize these 
facts serious difficulties have been encountered, and their 
discussion has not yet resulted in a satisfactory theory of 
the causes of change. All students of the subject feel the 
need of more extended observation, and from the point of 


view of the theorist there is special interest in the history 
of changes in regions remote from those which have here- 
tofore received chief attention. Comparatively little is 
known of the history of American glaciers, but the avail- 
able data have been carefully collated by Russell^ and 
Reid,^ and an important contribution has recently been 
made by Klotz.^ 

The local descriptions of the preceding pages contain 
the additional observations made by the Harriman Ex- 
pedition, together with inferences as to modern changes; 
but as the paragraphs on variations are somewhat scat- 
tered, the principal inferences are here assembled in re- 
sume. The geographic order, from east to west, is re- 

In Glacier Bay the observations pertain to a large area, 
containing at the present time a considerable number of 
separate trunk glaciers. The data concerning variations 
are not equally full in all parts, but so far as comparable 
they are harmonious. It is probable that the history of the 
whole district centering in the bay is a unit. The history 
begins with an epoch when the glaciers were smaller than 
now. During this epoch a forest grew to maturity and 
then was overwhelmed by gravelly waste from the ice; 
the epoch was therefore measured by centuries. The 
glaciers then advanced many miles, attaining a maximum 
one hundred or one hundred and fifty years ago, and they 
have since retreated. Measured to the Muir Glacier, the 
total retreat to 1899 was more than fifteen miles; meas- 
ured to the Grand Pacific, it was more than thirty-five 

1 Climatic Changes indicated by the Glaciers of North America : Am. Geol., 
vol. IX, pp. 322-336, 1892. Reprinted, with little change, as chapter viii of 
Glaciers of North America. 

2 Variations of Glaciers : Jour. Geol., vol. iii, pp. 278-288, 1895; vol. v, pp. 
378-383, 1897; vol. VI, pp. 473-476, 1898; vol. VII, pp. 217-225, 1899; vol. 
VIII, pp. 154-159, 1900; vol. IX, pp. 250-254, 1901 ; vol. X, pp. 313-317, 1902. 

'Notes on Glaciers of southeastern Alaska and adjoining territory. By Otto 
J. Klotz. Geog. Jour., vol. xiv, pp. 523-534, 1899. 


miles. The retreat was interrupted by a temporary ad- 
vance between 1890 and 1892. 

La Perouse Glacier is practically at maximum now. It 
has not been greater for centuries, except that it was a 
little longer a few years ago, when it invaded a mature 
forest. As its neighbors at the east and west were much 
smaller a century ago, analogy suggests that the present 
maximum was preceded by an important minimum. 

In Disenchantment Bay and its dependencies a great 
retreat, amounting to at least five miles along one channel 
of outlet and thirty miles along another, has been in prog- 
ress for more than a hundred years. There is evidence 
of other changes, but their order and dates are unknown. 

Columbia Glacier is practically at maximum now, and 
was nearly as large in 1794, but an important minimum 
probably occurred within the nineteenth century. It has 
not been greater than now for centuries, except that it was 
a little larger about the year 1892, when it invaded a 
mature forest. 

The numerous glaciers of Port Wells, including College 
and Harriman fiords, may have a harmonious recent his- 
tory, but the data are too meager to warrant a definite state- 
ment. They are somewhat smaller than at a maximum 
which may have occurred fifty to one hundred years ago. 

Grewingk Glacier has not for centuries been much 
larger than now. It was somewhat larger between fifty 
and one hundred years ago, and may have had a subse- 
quent maximum of nearly the same extent. 

The most conspicuous fact brought out by the compari- 
son of local histories is that they are dissimilar. Never- 
theless, there are limited resemblances. The Glacier Bay 
and Disenchantment Bay histories agree in including a 
great retreat, occupying more than a century. The Port 
Wells and Grewingk histories agree in a moderate retreat 
occupying something less than a century. The La Perouse 



and Columbia histories agree in a present condition of 
maximum glaciation probably preceded by an important 

As glaciers grouped together (about Glacier Bay, Dis- 
enchantment Bay, etc.) have seemed to vary in harmony, 
it is natural to look for a systematic geographic arrange- 
ment of the diverse histories; but such arrangement is not 
apparent. Port Wells and the Grewingk Glacier, inter- 
mediate in type of variation, are the most westerly of the 
localities (see map, fig. 55). Between Glacier Bay and 
ment Bay, rep- 
resenting one 
extreme of 
variation, flow 
La Perouse 
Glacier and its 
neighbors rep- 
resenting the 
opposite ex- 
treme. La 
Perouse and 
Columbia gla- 
ciers, agreeing 


in phenomena fig. 55. 

of variation, are separated by the contrasted phenomena 

of Disenchantment Bay. 

Glacier Bay adjoins Lynn Canal, being separated only 
by a mountain range, and some of the high neve fields of 
this range feed glaciers of both slopes. On both sides the 
glaciers are believed to be retreating, but the front of the 
Davidson of Lynn Canal is less than half a mile from the 
forest on its old moraine, and the Muir of Glacier Bay is 
nearly twenty miles from the equivalent forest. The great 
Fairweather Range, separating Glacier Bay from the Gulf 


of Alaska, nourishes many glaciers. Of these the Johns 
Hopkins, descending northeastward, has shared the great 
retreat of Glacier Bay; but the Brady, flowing south, and 
the La Perouse and Crillon and their neighbors, flowing 
southwest, have advanced during the same period. 

Close to Disenchantment Bay lies the Malaspina, a pied- 
mont glacier fed by alpine glaciers of the St. Elias Range. 
From other slopes of the same range come the principal 
feeders of the Hubbard, the main glacier of Disenchant- 
ment Bay. In a century or two the Hubbard has retreated 
five miles up Disenchantment Bay, but the Malaspina is 
bordered in places by a mature forest from which it has 
retired only a short distance, and at one point it has even 
advanced against the forest within a few years.^ 

The general fact appears to be that mere proximity 
does not ensure parallelism of glacial history; on opposite 
sides of a mountain range the sequences of change may 
be not only different but antithetic. 


In the discussion of the causes of the advance and 
retreat of European glaciers the phenomenon occasioning 
greatest difficulty is the lack of parallelism between the 
variations of different glaciers and different groups of gla- 
ciers. The histories of glaciers of the Alps exhibit dis- 
parities similar to those of Alaska, with the apparent 
difference that the Alaskan disparities are of larger scale, 
just as the Alaskan glaciation has a larger pattern; and 
the arctic and boreal glaciers of Europe probably exhibit 
equal irregularity, although Rabot, who has recently 
assembled the evidence, finds a number of partial corre- 

II. C. Russell. Amer. Geo!., vol. ix, p. 329, 1892. 

^Les variations de longueur des glaciers dans les regions artiques et bor^ales. 
Arch, des Sci. Phjs. et Nat., 4me periode, vols. 3, 7, 8, 9, Geneva, 1897-1900. 


For the explanation of irregularity there are two promi- 
nent hypotheses. The first, regarding oscillation of long 
period as climatic, ascribes minor oscillations to the 
rhythmic gorging and disgorging of the neve reservoir. 
The rhythmic period, being connected with topographic 
conditions, is different in different glaciers. 

The second, which has been more fully developed, ap- 
peals to the principle of lag. The chief cause of variation 
of the wasting end of the glacier is believed to reside in 
variation of snow accumulation on the neve fields, but con- 
siderable time is required for the transmission of the effect 
from end to end of the glacier. For average Alpine 
glaciers this time is believed to amount to several decades, 
but it varies with the length, slope, and other peculiarities 
of the individual ice streams, and the general result of its 
variation is that ice streams of the same mountain slope, 
or streams flowing from the same neve, initiate a period 
of advance or retreat in different years or even different 

It is impossible to compare the first hypothesis with the 
behavior of Alaska glaciers, on account of the meagerness 
of observational data. As yet we know nothing of perio- 
dicity of variation. The comparison of the second is ob- 
structed by the lack of meteorologic records for the glacier 
region, and by the fact that very few of the determinations 
of variations are associated with definite dates; no com- 
parison with climatic variation is possible, and the com- 
parison of individual histories of variation, one with 
another, is approximate only. Nevertheless, some judg- 
ment may be formed of the general competence of the 
second theory; and it seems to me not fully adequate for 
the explanation of Alaskan disparities. 

Consider, for example, the contrasted histories on oppo- 
site sides of Fairweather Range. On the northeast slope 
a great advance culminated not less than one hundred 


years ago, and a great retreat has been in progress ever 
since. On the south and southwest slopes the same period 
of one hundred years has witnessed a general advance. 
The century's variation for one side of the range, in a 
general way and so far as known, is the reverse of the va- 
riation for the other side. To apply the principle of lag to 
the phenomena it is necessary to suppose that the glaciers 
of one slope are very far behind those of the other in some 
phase of variation. If the southwestern group are con- 
sidered the slower to respond to variations of neve supply, 
the inference is that they are now at or near a culmination 
corresponding to the culmination of the northeastern 
group a century or more ago. If the northeastern group 
are considered the slower to respond, the inference is that 
though now at very low ebb, they have not yet felt the 
impulse which has carried the southwestern group to a 
maximum; and many decades, if not a full century, will 
be required to bring them to the same phase. It is not 
clear to me which horn of the dilemma should be taken, but 
in either case the time interval between corresponding 
phases is greater than can reasonably be ascribed to lag. 
Associated with the theory of lag as applied to the 
variations of European glaciers is a generalization that 
glacial variation is rhythmic, with a period of about thirty- 
five years, each recurrent cycle of variation being brought 
about by a corresponding cycle of climatic change. In 
this respect also the Alaskan phenomena are discordant. 
It is not credible that the great advance and retreat which 
occurred in Glacier Bay, involving the extension of glaciers 
along the main trough for thirty-five miles, or more, and 
their subsequent melting, could be accomplished in so 
short a period as thirty-five years; for, though direct ob- 
servation has covered but a small part of the great oscil- 
lation, it has shown that in the half of thirty-five years the 
retreat of the ice front was less than five miles. 


Reid has suggested a local subsidence of the land as 
a possible explanation of the retreat of the glaciers of 
Glacier Bay.^ Stumps of trees that grew in Muir Inlet 
before the great advance of the glacier, now stand at low- 
tide level, and demonstrate a submergence of at least 
twenty feet. The submergence may have been greater; 
and he points out that any lowering of the surrounding 
land with reference to the sea would make the conditions 
less favorable for the accumulation of snow and tend to 
reduce glaciers. To extend this explanation so as to 
cover the diversity of local histories it would be necessary 
to assume that the Fairweather Range was not lowered 
in company with the adjacent tract about Glacier Bay; 
and it would be logical also to assume that the great ex- 
pansion of Glacier Bay ice which preceded its shrinkage 
was associated with a rise of the surrounding land. As 
there is independent ground for believing that the region 
is one of active mountain growth, the occurrence of such 
differential and diverse movements is quite conceivable, 
and their possibility should be kept in view in the study 
of each locality. But as glaciers are highly sensitive to 
climatic changes, as climates are subject to continual and 
rapid variation, and as earth movements are comparatively 
slow and moderate in their influence, the central theory 
of glacier variation is necessarily climatic rather than 

With reference to the climatic explanation of the Alaskan 
phenomena I have a suggestion to contribute — a sugges- 
tion of a somewhat vague character, not yet reduced to 
the form of a definite hypothesis. It is, that the combi- 
nation of a climatic change of a general character with 
local conditions of varied character, may result in local 
glacier variations which are not only unequal but op- 

*Nat. Geog. Mag., vol. iv, p. 40, 1892. 


The general drift of the suggestion may be illustrated 
by considering some of the more evident consequences of 
an assumed change in the temperature of the water of the 
Gulf of Alaska. Let us assume that the water becomes 
warmer, and that all other factors affecting glaciation 
remain unchanged. The consequences would include: 

1. A higher temperature for the air currents flowing 
from the gulf to the land. 

2. A greater contrast in temperature between the coastal 
belt and the interior of Alaska, especially in winter. 

3. Greater evaporation from the ocean and a higher 
humidity for the landward-flowing air — resulting from i. 

4. Greater precipitation on the mountains, especially in 
winter — resulting from 2 and 3. 

5. A shorter annual period in which precipitation takes 
the form of snow — resulting from i. 

6. A (probably) lower ratio of snow to rain — resulting 
from 5, qualified by 4. 

7. A higher snow-line. 

8. More rapid waste of ice and snow by evaporation 
and melting — resulting from i, 5 and 7. 

Of these consequences, the increase of precipitation 
would tend to enlarge glaciers, while the lessened ratio of 
snow precipitation and the enhanced wasting would tend 
to reduce them. 

Evidently a lowering of the temperature of the gulf 
water would be followed by the reverse consequences. 

If the hypothetic rise of ocean temperature were carried 
to an extreme, the snow-line would be driven above the 
mountain tops and the glaciers would disappear. If the 
hypothetic fall of ocean temperature were carried to an 
extreme, so as to abolish the contrast between sea and 
land temperatures, the southern coast of Alaska would be 
reduced to the condition of the western coast, and glaciers 
would disappear from all but the highest mountains. 


There is, therefore, a temperature of ocean water which 
is more favorable for the development of glaciers in the 
coastal mountains than a higher or lower temperature. 
But this temperature is not the same for all parts of the 
coastal belt; it must vary with local topographic charac- 
ters. The glaciers of a low range may be more sensitive 
to variations of the snow-line than those of a high 
range. Glaciers facing the sea may be more sensitive 
to the variations of wastage dependent on the tempera- 
ture of the incoming winds than are glaciers facing the in- 
terior. Glaciers fed from cirques, where snow is concen- 
trated by wind and avalanche, may respond to variations 
of precipitation in a very different way from glaciers fed 
by open neve fields, where much of the annual snowfall is 
dissipated by dry evaporation. The laws of variation for 
high-grade glaciers may be quite different from those for 
glaciers of gentle slope. And so, when the ocean tem- 
perature approximates the value most favorable for the 
development of glaciers in the district as a whole, it will 
be too warm for the highest development of certain gla- 
ciers and glacier systems and too cool for others. And 
whenever such a condition obtains, a change in ocean 
temperature will cause some glaciers to enlarge and 
others to contract. 

It is of course impossible that one of the meteorologic 
conditions determining Alaskan glaciation should vary by 
itself while all other conditions remain constant, and the 
case assumed for the sake of illustration is therefore 
purely ideal. It has served its purpose if it has given 
plausibility to the suggestion that a change in some 
meteorologic factor or factors may result in simultaneous 
modifications of glaciers which differ not only in amount 
but in algebraic sign. 

Whatever may be the causes of the variations of gla- 
ciers, Alaska affords an inviting field for their investiga- 


tion. Intermediate in accessibility between the glacial 
districts of Europe and Greenland, it is more comprehen- 
sive than either in variety of local circumstance. The 
complexity of interacting conditions may for a long time 
baffle attempts at analysis, but when the complexity has 
been resolved, the resulting theory should have wider 
application than one founded on simpler phenomena. 



All geologists who have studied the region we tra- 
versed are agreed that the glaciers were much more ex- 
tensive in Pleistocene time than now; but opinions dif- 
fer widely as to the actual magnitude of the ancient ice 
fields, and also as to the extent to which they modified the 
topography of the country. As the nature of my journey 
rendered my view somewhat cursory and superficial, and 
as nearly all parts of my route had been covered, with 
better opportunity, by one or more of my predecessors, 
I can not expect to settle any of the vexed questions, but 
it still seems best to make rather full record of my obser- 
vations and impressions. When an observer views a 
complex phenomenon his attention is naturally directed 
to the particular features which his previous training en- 
ables him to appreciate — he " sees what he has eyes to 
see"; and the difference of eyes makes the work of 
independently trained observers more or less comple- 




Of the various classes of evidence from which the his- 
tory of Pleistocene glaciation is inferred, the physiographic 
is most available to observers who see the land chiefly from 
the deck of a vessel. Ice-scoured surfaces referable to 
the ancient glaciers were occasionally discovered during 
our journey, and a few drift deposits were closely exam- 
ined; but such observations served chiefly as checks on 
inferences from topographic form. 

The general characters of the physiographic data which 
may be used in such studies are familiar and need not be 
recited here, but a special sculpture feature — the hanging 
valley — may need introduction to some of my readers. 
Its utility in the interpretation, discrimination and estima- 
tion of the work of Pleistocene glaciers has been little 
appreciated until quite recently,^ but in the study of the 
Alaska field it was found extremely useful. 

A hanging valley is a small U-valley tributary to a 
larger valley, the floor of the smaller being considerably 
higher at the junction than the floor of the larger. Many 
of them are short, high-grade troughs, heading in cirques; 
some are mere cirques, without troughs — spoon-bowl hol- 
lows, high on the walls of main valleys. They are asso- 
ciated with other evidences of glacial sculpture, and the 
elevation of their floors is believed to result, as a rule, 

1 Lake Chelan, by Henry Gannett : Nat. Geog. Mag., vol. ix, pp. 417-428, 1897. 
Glacial erosion in the valley of the Ticino, by W. M. Davis : Appalachia, vol. 
IX, pp. 136-156, 1900. Glacial erosion in France, Switzerland and Norway, by 
W. M. Davis: Proc. Boston Soc. Nat. Hist., vol. xxix, pp. 273-322, 1900. 
Review of the last by T. C. Chamberlin : Jour. Geol., vol. viii, pp. 568-573, 

Davis's second paper reviews the literature, points out that McGee, De Lap- 
parent and Richter had advanced somewhat similar ideas as to the origin of 
hanging valleys before the appearance of Gannett's paper, and mentions an 
unpublished address by Penck. The glacial explanation of the hanging valleys 
of the Alps is opposed by Bonney and Garwood in Quart. Jour. Geol. Soc. 
London, vol. LViii, pp. 690-718, 1902. 



from the unequal erosion of valleys by glaciers of un- 
equal size. 

Where a trunk glacier of alpine type receives a rela- 
tively small lateral tributary the main ice stream gives 
base-level to the tributary. Their relation in this respect 
is homologous with the relation of main and tributary 
streams of water. If the magnitude of trunk and tribu- 
tary have remained constant long enough for erosion to 
bring about an adjustment of grades, the surface of the 
tributary at the point of junction has the same level as the 
adjacent surface of the trunk. But the bottoms of the 
two channels do not have the same level. The trunk 
stream is normally deeper than the tributary, and at their 
junction its bottom lies at 
a lower level. If the 
streams (of ice or water) 
be removed, the bottom of 
the tributary channel is 
found to end high up on 
the side of the main 

The ideal case, diagram- 
matically illustrated in 
figures 56 and 57, is also il- 
lustrated in the actual topo- 
graphy of many regions ^,^,^ ^^^^, ^^, diagrams illustra. 
sculptured by Pleistocene ting origin of hanging valleys. 

1 • 56. A terrestrial block containing a trunk 

giaCierb. glacier and tributary with well adjusted chan- 

The hanging valley is es- ^^eis. 

, , , 57. The same block without the ice, show- 

pecially significant in two ing the adjusted glacier channels. The trunk 

.. /• T_ • l,* • channel is deeper than the tributary; the trib- 

ImeS 01 physiographic in- utary channells a hanging valley on the side 

terpretation. It is a con- of the trunk channel. 

spicuous earmark of the former presence of glaciers ; and 
it helps to a conception of the magnitude of Pleistocene 
glacial erosion. 


Its value as an earmark depends on the principle of 
exclusion; glaciation is the only physiographic process 
known to produce such forms. It is true that discordance 
of level between trunk and tributary valleys is not by itself 
diagnostic of glaciation, for it often occurs as a temporary 
condition in systems of stream-made valleys, especially 
when fresh uplift stimulates down-cutting by trunk 
streams; but in such cases discordance is associated with 
the narrow trenches of youthful or rejuvenated topog- 
raphy. It is true also that the glacial U-trough is some- 
times (though rarely) simulated by products of stream 
erosion, and that a hollow closely resembling the glacial 
cirque is occasionally produced by aqueous process; but 
these imitative forms belong to the middle life of a stream, 
when down-cutting has so slackened as to permit valleys 
to broaden, and they imply a harmonious grading of stream 
beds, inconsistent with discordance of level at the junc- 
tion of tributary and trunk. But the combination of dis- 
cordance of level with U-shaped cross-profiles constitutes 
a physiographic type peculiar to the work of glaciers. 

The significance of the hanging valley for the valuation 
of glacial erosion depends largely on the assumption that 

the discordance of level was 
produced by the glacial exca- 
vation of the main trough, and 
this assumption requires qual- 
ification. If A^C in the dia- 
FiG. 58. DIAGRAM ILLUSTRATING gram (fig. 58) bc thc cross- 


profile of a mam glacial trough, 
DjE the longitudinal profile of a tributary trough, and JSJF 
the produced floor of the tributary, the discordance' is rep- 
resented by the height of J^ above ^. If the preglacial 
stream valleys were accordant in grade, their junction 
was at jFf or some point above jF; and J^jB is the meas- 
ure either of the deepening of the main trough by the 



•>n, tor 

times (th( 





A Hanging Valley, Eraser Reach 

Eraser Reach is a long, deep channel, or fiord, of the ' Inside Pas- 
sage,' here about two miles wide. The hanging valley is in Princess 
Royal Island, and joins the fiord trough from the southwest. Its sill 
is several hundred feet above the water surface. 

Erom a photograph made by E. S. Curtis, June 3, 1S99. Negative 
NO. 187. 


^t of 7 7?. T 

Ui ihc aiain trough by 

H.A.E. VOL.111 



A "Hanging ValleyJ Fraser. Reach 


glacier, or of the difference between that deepening and 
the deepening of the tributary. If the preglacial val- 
leys were discordant, the main valley must have been 
young, and was probably so narrow that its conversion 
into a U-trough involved much excavation at the sides. 
In either case, therefore, the occurrence of high hanging 
valleys is indicative of great erosion by ice. 

Yet another qualification is necessary, for there is a 
special condition under which a relatively young stream 
valley may have the width of a glacial U-valley. When 
the activity of streams is revived by uplift, the general 
degradation may be outstripped not only by trunk streams 
but by strike streams following outcrops of weak rocks ; 
and if the contrasts of rock resistance are great, a broad 
belt of weak rock may induce the development of an open 
valley or wide trough, while neighboring uplands of re- 
sistant rock are still little modified. Thus may arise a 
topographic condition susceptible of conversion by only a 
moderate amount of subsequent ice work into a typical 
glacier trough with hanging valleys. 

Theoretically, then, there are at least three cases to be 
borne in mind in inferring the quantity of glacial erosion 
from the existence of hanging valleys. First, the grades 
of preglacial streams may have been accordant, in which 
case the discordance of hanging valleys with trunk valleys 
yields a rough measure of the depth and amount of glacial 
erosion. Second, the grades of preglacial streams may 
have been discordant, and without dominant control of 
contrasted rock texture, in which case the erosive work 
of ice may have consisted chiefly in the enlargement of 
V-gorges to U-troughs. Third, the discordance of pre- 
glacial grades may have been associated with the rapid 
opening of valleys in weak rocks, in which case the erosive 
work of ice may have been small. The criteria for dis- 
criminating the three cases have not been worked out. 


but a few general propositions may be advanced. In the 
first case a study of the district should discover independ- 
ent evidence of the maturity of the preglacial topography. 
In the second and third there should be independent evi- 
dence of rejuvenation of preglacial streams. The third 
could not arise unless the main troughs follow the strike. 

The practical problem is further complicated by the fact 
that the type of initial topography, when determined, leads 
to only a limiting value for the total erosion, and also by 
the complexity of glacial history as dependent on climatic 
variation. A glacier which begins erosive work by 
broadening and rounding the cross-profile of a stream 
gorge does not cease activity when that result is attained. 
One which falls heir to a weak-rock strike valley, fairly 
adjusted to its conditions of flow, may carry the work of 
excavation far beyond the grade limit of the ancestral 
stream, and hollow out a lake basin or fiord trough. And 
the coordinated system of grades and channel forms toward 
which the erosive work of grouped glaciers tends, is itself 
modified by every change of the general volume of ice. 

But despite all qualifications the hanging valley is the 
most important witness yet discovered to the magnitude 
of the work accomplished by the alpine glaciers of the 

The hanging valleys of Alaska are illustrated by many of 
the figures and plates of this volume. The mouth of one 
overlooking Hidden Glacier is imperfectly shown in plate 
V and figure 28, and an alpine valley truncated below by 
erosive action of the glacier appears in plate vi. Figure 
30 shows the mouth of a hanging valley above Nunatak 
Fiord; a glacier issuing from a hanging valley north of 
Nunatak Glacier is shown in figure 31 ; and a glacier cas- 
cading from a hanging valley of the south side of the 
same trough in figure 32. Figure 43 shows a tributary to 
Yale Glacier issuing from a hanging valley, and figures 44 


and 45 a series of similar valleys and glaciers bordering 
College Fiord and Harvard Glacier. In figure 46 two 
hanging valleys are shown, the one empty, the other fur- 
nishing a tributary to Barry Glacier; in figure 50 are two 
high valleys with small glaciers overhanging Serpentine 
Glacier; and figure 51 represents Cataract Glacier, issuing 
from a high valley and cascading to Harriman Fiord. All 
these examples, occurring in the preceding portion of this 
report, are incidentally included in views selected to illus- 
trate other features. In the following portion are a num- 
ber of views chosen wholly or partly with reference to 
hanging valleys ; and the remark applies especially to plate 
XVI and to figures 62, 66, 69, 70, 71, 72, 74, 76, 77, 81, 88 
and 93. 


From Puget Sound, Washington, at the south, to Lynn 
Canal and Glacier Bay, Alaska, at the north, a space of 
900 miles, the coast of North America has a peculiar and 
significant facies. It is divided into a fringe of rugged 
peninsulas by deep, narrow inlets, and guarded from the 
surges of the open ocean by a great number of rocky 
islands and islets. In this respect it resembles the coast 
of Maine and the western coast of the Scandinavian pen- 
insula, and, like them, its peculiar characters are associ- 
ated with evidences of extensive glaciation. It differs 
from those coasts in the fact that some of its islands are 
of great extent, so as to include or be constituted by 
mountain ranges, and in this respect it is paralleled by a 
single district only, the western coast of the southern ex- 
tremity of South America. 

The map of the district, figure 59, though drawn to so 
small a scale as to show only the larger islands and prin- 
cipal fiords, serves to illustrate the intricate penetration of 
the land by the sea. It is reduced from the large chart 
(3689) of the U. S. Coast Survey. 






The general trend of the coast of the mainland is north- 
west to north-northwest, and this trend is shared by the 
longer axes of the principal islands. There are thirty-two 
islands exceeding twenty miles in their greater dimensions, 
the maps show more than 400 islands above one mile in 
extent, and the islets are uncounted. 

With minor exception, the peninsulas and islands are 
mountainous, descending steeply to the water, and the 
passages between them are deep. Most of the inlets of 
the mainland 
and many of 
the passages 
dividing islands 
are of approxi- 
mately uniform 
width for long 
distances, and 
the parallel 
shores of such 

linear water- ^'^' ^* ^^ alaska fiord ; tracy arm, holkham bay. 

ways are usually steep and somewhat simple in contour. 
Repeating thus the characteristic features of the Norwe- 
gian coast, they fall within the physiographic class to which 
the name fiord is applied. The northwesterly trend char- 
acteristic of the islands affects also the passages between 
them, and has enabled navigation to select for its use a route 
close to the mainland, where deep waters are almost wholly 
protected by the islands from waves and storms of the open 
ocean. This route is commonly known as the * inside pas- 
sage.' Our steamer followed it on both the outward and 
return voyages, our principal deviation from it being 
made by a visit to Sitka, which lies on the southwest or 
oceanic side of Baranof, one of the outer line of islands. 
Parts of the journey were made at night, but the outward 
and inward courses, taken together, showed us much the 


greater part of the inside passage, and our examinations 
about Sitka occupied several days. An excursion to 
White Pass gave a view of a canyon at the head of a 
fiord, and of uplands at 2,000 to 2,500 feet altitude, but 
the higher summits were seen only from stations at or 
near sea-level. 

Direct observation of the uplands was afterward supple- 
mented by the study of photographs. The U. S. Coast 
and Geodetic Survey and the Canadian International 
Boundary Commission have used the camera freely in con- 
nection with topographic surveying, and I was so fortu- 
nate as to have access to their series of pictures. The latter 
organization spread a web,of triangulation over all the 
mainland portion of southeastern Alaska, and from each 
of the mountain peaks occupied as stations photographed 
the entire horizon, using the views afterward for the con- 
struction of contour maps. Their album thus represents 
the upland in a systematic and thorough way and is emi- 
nently adapted to physiographic study. (See page 6.) 

In this, as in other districts of Pleistocene glaciation, it 
is evident that the Pleistocene sculpture is superposed on 
an earlier sculpture, chiefly aqueous; and in the discus- 
sion of the work of Pleistocene glaciers it is necessary to 
consider the pre-Pleistocene condition. I find it conve- 
nient to begin with that consideration. 

Pre-Pleistocene Tocography 
So far as we saw the indurated rocks, and so far as we 
know them from the descriptions of others, they are either 
igneous or metamorphic. The igneous rocks are almost 
wholly intrusive. The metamorphic exhibit various de- 
grees of alteration, but all are so folded or squeezed that 
the planes of structure make large angles with the hori- 
zon. The general strike is believed to be parallel to the 
coast, but there are few direct observations of strike. In- 


ference rests partly on the trends of structure lines deter- 
mined by the Canadian Geological Survey in neighboring 
parts of the continent, partly on the trends of the straighter 
valleys and channels of the region itself. 

From this complex the pre-Pleistocene topography was 
developed by erosion. The only constructional forms we 
saw which might have antedated the Ice Age were a few 
volcanic cones. The system of relief was related to three 
known base-levels. The plane of the first is now high in 
air, above some of the mountains and among the peaks of 
others. The second is not far from present sea-level, and 
the third is below sea-level. 

The High Peneplain, — The uplands of the mainland 
are remarkably uniform in general height over large areas, 
not indeed presenting plain surfaces, but either exhibiting 
harmony of crest lines, despite profound and general dis- 
section, or else occupied by numerous small shallow val- 
leys, which are strongly contrasted with the deep steep- 
walled trenches of a less complete dissection. These 
features can be most readily presented in connection with 
some of the accompanying illustrations. Figures 3, 61, 
62, 6t^^ 75, and 77 were drawn from photographs by the 
Canadian Boundary Commission. 

Figure 61 shows the upland topography north of the 
western end of Cross Sound. (The reader can identify 
the locality on the map, page 120, as the third cape west 
of the mouth of Glacier Bay.) We stand on a summit 
above Cape Spencer, and look northwest. At the left is 
the Pacific Ocean; in the center distance, the end of 
Fairweather Range (the nearest high peak being La Pe- 
rouse, 10,750 feet) ; at the right, Brady Glacier, its foot 
separated from Taylor Bay by a gravel strand. Between 
us and the base of the mountains, 18 miles away, are a 
series of hills somewhat uniform in height. The higher 
points (as we learn from the Commission's contour map) 


range from 2,000 to 2,500 feet above tide, and near the 
mountain rise above 3,000 feet. 

Viewing these hills collectively, one can hardly fail to 
be impressed with the appearance of system in their 
crest lines. Some of the hills have broad and flattish, or 
gently arched, backs; and all the higher parts of their 
profiles seem to belong to a single gently sloping plane. 
Should the valleys between them be filled up even with 
the crest lines, the group would become a plateau with 
undulating surface. It is natural for the geologist, when 
he sees such a harmonious arrangement of hill tops, to 
seek an explanation in the structure of the rocks, but in 
this case a structural explanation can not be found. The 
rock layers are not horizontal but nearly vertical, and 
they have been cut across in the shaping of the land. 
The local elements of the upland forms are due purely to 
erosion. The most probable explanation of the phenom- 
ena is that the area was first worn down to a plain at or 
near sea-level, afterward raised so as to be a plateau, and 
then dissected into a group of hills. The habit of the 
hills indicates that the principal work of dissection was 
by streams, but there was also glacial sculpture. They 
were overrun by the ice-sheet, and the glacial rounding 
of summit angles helped to obscure, though it failed to 
destroy, the evidence of the old base-level plain. 

Yet other evidence of the geologic history is connected 
with the trends. The rock structure strikes northwest, 
or from the foreground toward the mountains, and this is 
also the trend of the upland as a whole. But the crests 
of individual hills trend east of north, making angles of 
50° to 60° with the strike; and the separating valleys 
have the same trend. The valleys do not head against 
high summits among the hills, but traverse the plateau 
from side to side. They seem to be the work of a system 
of streams whose courses across the plateau were deter- 



^ o 

^ .2 

mined by some 
cause other than 
the rock struc- 
ture. As their 
direction coin- 
cides with that 
of the local in- 
clination of the 
old peneplain, it 
is highly proba- 
ble that they 
were super- 
posed on the 
rock structure at 
the time of the 
uplifting of the 
plateau. The 
great valley 
holding Brady 
Glacier and Tay- 
lor Bay, on the 
other hand, runs 
parallel to the 
strike of rocks 
exposed in the 
hills, and may 
fairly be as- 
sumed to occupy 
the outcrop of a 
belt of compara- 
tively weak 

If now we 
project the sum- 
mit plane of the 


hills backward across the Brady-Taylor valley, we find 
that it passes among the highest peaks of another and 
higher upland, an upland lacking the broad summits of the 
Cape Spencer hills, but characterized instead by notable 
uniformity in the height of numerous acute summits. A 
suggestion of this character may be seen at the right 
in figure 6i. 

Before leaving this view, note should be made of the 
fact that the peneplain of the Cape Spencer hills ends 
northward at the base of Fairweather Range. The range 
is distinct in physiographic type and in geologic history. 

The reader is now asked to turn back to figure 3, drawn 
to give a bird's-eye view of Davidson Glacier, but show- 
ing also the east wall of Lynn Canal. The point of view 
is not quite so high as the crest of the opposite wall, but is 
high enough to show that the upland bounded by that wall 
has plateau characters. There are none of the smooth 
crest lines seen in figure 61, but angular peaks and crests 
standing close to the mural face combine with angular 
peaks and crests farther back to give an even sky-line. 
All the valleys visible are upland valleys, and it is not 
hard to believe that these have been carved out of an 
uplifted block, the plane of whose original flat top runs 
among or above the phalanx of sharp summits. 

This upland stands about 3,000 feet higher than the 
hills of Cape Spencer, and is eighty miles northeast of 
them. The intervening uplands are parted by two great 
fiords, Lynn Canal and Glacier Bay, and are considerably 
dissected in detail, but where most massive they have the 
plateau habit shown in figure 3 ; and they are intermediate 
in height between the plateaus at east and west. 

In figure 62 we have the view commanded from a peak 
like one of those against the sky in figure 3. The locality 
is twenty-five miles farther south, and we are looking 
eastward from a point about five miles east of Lynn Canal. 


The higher summits have a general altitude of 5,200 feet. 
It is now evident that the sculpture is that characteristic 
of the work of local glaciers. The sharp peaks overlook 
cirques, many of them still filled with ice; below the 
cirques are short glacier troughs ; and between the troughs 
are narrow crests. These glacier troughs are not insig- 
nificant features — the beds of those in the field of view lie 
1,000 to 2,500 feet below the plane of the high peaks — 
but they are so small in comparison with the great earth 
block from which they have been hewn that they do not 
prevent the imagination from restoring its original outlines. 


A deep long glacier trough traverses the view from dis- 
tance to foreground. It is occupied in part by a glacier, 
in part by a filling of glacial waste; and its rock bed is 
far below sea-level. It is related to the fiords and must 
be considered in another connection. In the present con- 
nection it need be thought of only as a trench dividing 
the upland, and helping, through contrast, to exhibit the 
upland's plateau character. 

Other illustrations of the plateau dissected by Lynn 
Canal and its branches are to be seen in figures 75 and 


77. In each case the point of view is so low that the up- 
land peaks do not unite in an even sky-line, but other 
plateau features are brought out. 

The correlation of the Cape Spencer peneplain with the 
Lynn Canal plateaus brings together very different topo- 
graphic types, but they are not essentially incongruous. 
The greater height of the inland district has caused it to 
be occupied by local glaciers, which have scooped out a 
system of cirques and rounded valleys, leaving the inter- 
vening crests angular. The cape district is and has been 
too low to initiate glaciers, but has been overflowed by 


an ice-sheet originating outside it. It therefore lacks 
cirques and the sharp crests developed by cirque erosion, 
but has suffered a somewhat equable reduction of its fiat- 
tish summits. 

In turning attention now to figure St^j we leave the 
region of Lynn Canal and pass three hundred miles south- 
ward to the Walker Bay region near Behm Canal. The 
general height of crests is here 4,000 feet. The even sky- 
line includes a few sharp peaks, suggesting the cirque 
sculpture of the northern area, but most of the distant 
summits and all of the nearer are rounded, as by an over- 


riding ice-sheet. The deep dissection indicates a subse- 
quent history different from that at the north, but the 
evidence of an initial peneplain is at least equally clear. 
The approximation of the summit heights to uniformity is 
too close to be accounted for without the hypothesis of an 
uplifted plain; but the departures from uniformity indicate 
that little if any of the original plain survives. 

The general interpretation of these upland features ap- 
pears to be as follows: After the folding and squeezing 
of the metamorphic rocks, there was a long period of 
erosion, in which broad tracts of the land were worn 
down nearly to sea-level. Then came uplift, producing 
a plateau from 3,000 to 6,000 feet high, and erosion has 
followed. In some places, at least, this plateau sloped 
gently toward the sea, and its plane may have remained 
everywhere continuous, diversified only by moderate 
flexures; but there is also possibility that it was inter- 
rupted here and there by faults. The period of subsequent 
erosion has been long enough for the development of local 
peneplains at a lower level, and in that time the plateau 
has been greatly modified. Not only has it been dis- 
sected by the eating out of gorges and valleys, but its 
back has been worn and fretted, largely by local glaciers, 
until all the original surface and much of the original form 
have disappeared. What remains is chiefly a tendency 
to uniformity of crest height in the ridges and peaks of 
certain districts. 

It is probably true also that vestiges of the high pene- 
plain are conspicuous only where the rocks are com- 
paratively durable. Our best examples are along the 
mainland east of the archipelago, and that region, accord- 
ing to Dawson, is largely granitic. The dominant rocks 
of the archipelago are metamorphic, and we saw little 
from the ship's deck to suggest that the uplands of the 
islands have a plateau habit. The mountains next the 


water do not in general rise very nearly to the plateau 
plane, and we were frequently able to see that those 
farther back are higher. Even where the channels 
between islands are most sharply incised it is probable 
that they are excavated in the bottoms of broader hollows. 

My general conception of the configuration at the date 
of the lower peneplain — a conception necessarily vague, 
as well as provisional — is that it included all phases of 
the topographic cycle, being infantile to adolescent where 
the rocks are most resistant, adolescent to mature in the 
greater part of the Alexander Archipelago, and senile 
where the rocks are weakest. In the granites were nar- 
row gorges, set far apart, and reduced to low grade only 
where draining large tracts. Between them were tabular 
uplands with mild relief, except for occasional unreduced 
peaks, or monadnocks. In the stronger metamorphics a 
system of graded waterways divided the upland into 
mountain ridges, with scattered remnants of the summit 
plain. The master streams were largely consequent to 
the seaward slope of the old plain, but were in part 
diverted to lines of strike; and minor streams were ad- 
justed to rock structure. 

Loiv Peneplains, — Along the passages and channels 
we traversed, bold coasts are the rule and forelands of 
any character the rare exception, but at two localities we 
saw unmistakable traces of a peneplain between mountain 
base and the descent to deep water, and in the light of 
their evidence it seemed proper to give a similar interpre- 
tation to various features of obscurer character. 

Annette and Gravina islands lie next to the mainland 
in the southern part of the Alexander Archipelago. Out- 
side them on the west is the great Prince of Wales Island, 
from which they are separated by a broad channel, 1,500 
to 1,700 feet deep. Each island contains a mountain mass 
bordered on the west by a low foreland, the foreland being 


related to the mountain as peneplain to monadnock. On 
Annette Island, where we landed, the foreland is not 
continuous, being divided by a bay, but its parts, together 
with several low islands, appear to be remnants of a more 
extensive plain. The rock is slate and mica schist, and 
that of the adjacent portion of the mountain quartzite. In 
detail the surface is somewhat uneven, low moutonnee 
bosses alternating with hollows that hold pools and 
bogs, but there is a general and gradual ascent from the 
sea front to the mountain base, where the altitude may be 
three or four hundred feet. Figure 64 presents one rem- 


Showing relation of foreland (peneplain) to mountain, as seen from New Metlakatla. 

nant of foreland in profile, as seen from the other, and in- 
cludes also rocky islets. 

As this whole region was deeply buried by Pleistocene 
ice, the unevenness of the foreland is readily understood 
as the result of glacial erosion subsequent to the original 
planation. Probably none of the original surface remains, 
but the glacial degradation must have been locally quite 
moderate, or the general plain character would not have 
survived. The phenomena do not yield a close determi- 
nation of the pre-glacial relation of sea-level to land, but 
it could not have differed greatly from the present. 

The second locality is Sitka Sound, 175 miles to the 
northwest, and on the ocean front of the archipelago. 
Back of Sitka the mountains of Baranof Island rise 
abruptly to a height of several thousand feet, and they 
are penetrated by inlets and lake valleys exhibiting a 
moderate development of fiord characters. But the town 



'■^1't:>H^' -,■ 

itself stands on a foreland carved from the rock, and this 
foreland slopes gradually under the water of the sound or 
bay. In detail the foreland is even more rugged than that 
of Annette Island, and where it passes beneath the water 
its eminences give rise to a great number of islets, which 
stud the sound and form the natural breakwater of Sitka 
Harbor. The relations of mountain, foreland and islands 
are well shown in plate xvii (lower view) and figure 
65, which represent Cape Baranof, a few miles south of 

Sitka. Here 
also the in- 
dicated base- 
level has ap- 
the height 
of modern 

and Wran- 
gell Strait, a region on the landward side of the archi- 
pelago, we saw more extensive tracts which probably 
pertain to the same base-level. They stand somewhat 
higher, averaging several hundred feet in altitude; and 
the parts we saw best have been so modified by glacial 
erosion that original base-leveling might not have been 
inferred without the aid of the Annette and Sitka exam- 
ples. They are illustrated by the upper view in plate 


Near the south end of Lynn Canal, Douglas Island is 
separated from the mainland by a narrow fiord, the Gas- 
tineau Channel. Facing the channel, the island is flanked 
by the ruins of a high rock terrace (fig. 66), A dozen 
short valleys of the island join the fiord at the level of the 
terrace, which descends southeastward from an estimated 

Seen from the timber line on Mount Verstovia. 

i /. ri iO /lOITA 

-do ofiJ fiv/(»} nir. . 
bniiloio'i ■//<. 'iji; 

jifc jiodix;ii :>.. i>njj 

ifKJfiuofn till J iiilq 

.lohj;?:-,' : 'jticliJ^. niiilqv ilud 


Upper Figure. — Lowlands near Wrangell 

Standing south of the main town the observer looks southwestward 
across the harbor. The narrow foreland on which the town stands, 
and the peninsula beyond the harbor, are composed of metamorphic 
rocks, and are probably remnants of a dissected and greatly worn pene- 
plain. • The rounded crests of the mountains beyond suggest that they 
were overridden by the Pleistocene ice-sheet. See page 132. 

From a photograph made by E. S. Curtis, June 5, 1899. Negative 
NO. 195. 

Lower Figure. — Foreland and Islands near Sitka 

The view looks south and southwest from the hillside back of Sitka. 
The islands of the harbor, and the foreland beyond. Cape Baranof , are 
believed to be parts of a peneplain subsequently worn by a glacier. 
See page 131, and compare figure 65, which gives a bird's-eye view 
of the same features. 

From a photograph made by E. S. Curtis, June 16, 1899. Nega- 
tive NO. ^3^.. 

Th< .r^ by the . 


•inland by 

IS tiaii 




altitude of 
1,000 feet to 
500 feet or 
less. The 
relation of 
this terrace 
to the drain- 
age suggests 
that it is a 
remnant of 
a pre -glacial 
valley floor, 
and if so it 
may be a 
grade plain 
ry with the 
low pene- 

These fea- 
tures point to 
a long con- 
tinuance of 
base-level at 
approxi m- 
ately its pres- 
ent height, 
the time be- 
ing subse- 
quent to the 
uplift of the 
older pene- 
plain. It was 
during this 


period that the chief pre-glacial erosion of the plateau 
was accomplished. While the granite tables were being 
dissected and the harder metamorphics worn into varied 
and rugged mountains, flat valleys and plains were devel- 
oped from the broader bodies of weak rocks. 

A Lower Base-leveL — The degradation of the troughs 
in which lie the channels, passages and straits among the 
islands, and the fiords of the mainland, has been carried 
far below the horizon of the lower peneplains. Close to 
the Annette Island peneplain Clarence Strait has a depth 
of 1,675 ^^^^? ^^^ 5^ miles farther north, where the 
strait is narrowed by Cleveland Peninsula, a depth of 
2,100 feet is recorded. Thirty miles inland from Annette 
Island Behm Canal is 1,800 feet deep; and 60 miles in- 
land, toward the head of Portland Canal, is a depth of 1,250 
feet. The greatest recorded depth of Chatham Strait, 60 
miles from either end and 25 miles inland from the Sitka 
peneplain, is 2,900 feet; and its northward prolongation, 
Lynn Canal, has one sounding of 2,475 ^^^^' These fig- 
ures are selected from charted soundings which indicate 
great irregularity of bottom configuration; they are 
maxima, and not averages; but the averages also are im- 
pressive. In the main part of Lynn Canal (55 miles long) 
the average sounding along the line of greatest depth is 
1,300 feet. The similar average for the surveyed part (60 
miles) of Chatham Strait is 2,000 feet; for Stevens Passage 
1,000 feet, Frederick Sound 900 feet. Summer Strait 
1,100 feet, Clarence Strait 1,500 feet, Behm Canal 1,400 
feet, Portland Canal 1,000 feet. These are all in the 
region of the Alexander Archipelago. The broad sound 
east of Queen Charlotte Islands, called Hecate Strait, 
ranges from 150 to 600 feet, and the narrow passages 
east of it are somewhat deeper. Queen Charlotte Sound, 
northeast of Vancouver Island, has a general depth of 
600 feet and a maximum of 1,140; the Gulf of Georgia a 


rough average of 900 feet and a maximum of 1,470; the 
narrow passage joining the sound and gulf an average of 
600 feet, with a greatest depth of 1,000. At the extreme 
south Puget Sound has an average depth of 500 feet and 
a maximum of 925. 

With the possible exception of Hecate Strait, these 
water-filled valleys are clearly products of erosion; and 
it is probable, though not proved, that they are younger 
than the low peneplains. Whatever the extent to which 
they were hollowed out by rivers, they were afterward 
greatly modified by glaciers; and the glaciers are respon- 
sible for the conspicuous unevenness of their floors. The 
demonstration of ice work is found in the thorough glaci- 
ation of all bordering lands, to be presently described, and 
in the sculpture of rocky islets, which have characteristic 
moutonnee forms. Much of the unevenness, and espe- 
cially the deeper basins, must be ascribed to glacial 
erosion, but a share may also be referred to glacial depo- 

Whatever the extent to which the hollows were deep- 
ened and enlarged by glaciers, it is probable not only 
that they were initiated by rivers, but that some of the 
rivers sunk their beds considerably below present tide- 
level. The most satisfactory evidence on this point was 
found at the extreme south. In the Puget Sound region, 
as shown by Willis, the ice movement was southward, a 
lobe of the ice-sheet ascending the broad valley of western 
Washington. This lobe made extensive modification of 
the face of the country, but chiefly by deposition and only 
secondarily by erosion. The system of troughs it left 
behind are regarded as preexistent stream valleys, only 
moderately scoured and straightened by the ice which 
overran and occupied them. ^ 

1 Drift Phenomena of Puget Sound. By Bailey Willis. Bull. Geol. Soc. 
Amer., vol. ix, pp. 111-162, 1898. 


This interpretation by Willis seems to me reasonable. 
The glacial deposits in the southern part of the Puget 
Sound region are voluminous, and much of their material 
is of distant origin. This is the place where the ice lobe 
discharged its load, and it is not probable that in the field 
of deposition the ice also developed by erosion a system 
of narrow deep troughs. Regarded as stream valleys, the 
channels of the sound tell of a pre-glacial base-level at 
least 500 feet, and probably 1,000 feet or more, below the 
present sea surface. 

A few features seen at the north might be regarded as 
confirmatory, but their interpretation is subject to con- 
siderable doubt. They are apparent exceptions to the 
general rule that at all low levels the sculpture forms of 
trough walls are glacial. 

But while the existence of a pre-glacial low base-level 
is on the whole probable, its precise relation to present 
base-level and the period of its duration are altogether 
conjectural. To bring all parts of the deep channels 
within reach of stream erosion it would need to be 3,000 
feet below present sea-level in the region of the Alexander 
Archipelago. Under present climatic conditions, such a 
change would carry a very large area above snow-line, 
and would so promote the alimentation of glaciers as to 
flood the whole district with ice and abolish stream ero- 
sion. Stream erosion, therefore, could not have been 
carried, by lowering of base-level, to the lowest parts of 
the channel system without the aid of important climatic 
variation. Without doubting the possibility of wide range 
in independent climatic factors, it seems easier to assume 
that the lowering of base-level was comparatively mod- 
erate, and that a considerable part of the down-cutting of 
the channels was accomplished by Pleistocene glaciers. 

There is equal doubt as to the duration of the low 
base-level, or the extent of the erosive work it enabled 


the streams to accomplish. The questions whether the 
broader channels of the archipelago were merely outlined 
by river gorges or were widely opened, whether the low 
peneplain was only trenched or was largely replaced by a 
peneplain at a level now submerged, whether the grading 
of river beds was restricted to the coastal district or was 
carried far into the interior, are not answered by any facts 
in my possession. 

To escape the confusion arising from the glacial re- 
modeling of water-wrought topography, it is natural to 
turn to the region just outside the glacial district. This 
region may be assumed to have shared the same oscilla- 
tions of base-level, so that whatever history may be derived 
from it can be transferred to at least the neighboring parts 
of the glacial district. The glaciation of what may be 
called the inner coast has its southern limit in Puget Sound. 
As the Olympic Mountains, separating the sound from the 
outer coast, contained Pleistocene glaciers, the outer coast 
also may have been modified by ice in that latitude. But 
farther south the coast was not directly affected by glacial 
ice. Between the Olympics and the mouth of Columbia 
River are two shallow bays, partitioned from the ocean by 
sand spits. The more northerly. Gray Harbor, receives 
the Chehalis River, a stream of moderate size, but the 
map gives no indication of a delta. Willapa Bay receives 
two small streams, Willapa and Nasal rivers, and these 
also are without deltas. It is evident that the bays are 
estuaries, or submerged portions of the river valleys, and 
they indicate a recent change in the relation of sea and 
land, the sea rising or the land sinking. 

Columbia River also ends in an estuary, its banks grad- 
ually separating, until near the sea they are ten miles 
apart. The estuary is shoal, and it is a matter of observa- 
tion that the parts protected from the current are being 
rapidly filled by the abundant silt of the river. The banks 


are in part low, but include also hills and bluffs. While 
this estuary is much larger than those of Chehalis and 
Willapa rivers, it is small in relation to the Columbia 
River, which carries a great volume of water, and the val- 
ley whose submergence it records was of very moderate 
dimensions. This valley was of course formed while the 
river ran at a lower level, but the erosive work accom- 
plished at that level was surprisingly small. As the con- 
tinental shelf is narrow along this part of the coast, the 
river may be supposed to have promptly graded its chan- 
nel to harmony with the depressed base-level, and the 
conditions would seem to have been favorable for the 
development of a broad and branching valley like that 
submerged in Chesapeake Bay. The fact that no such 
development took place seems to indicate either that the 
lowering of base-level was small or that the period of low 
base-level was short. Despite the great volume of the 
river, the valley developed by the discharge at lower level 
was quite insignificant in comparison with the fiords and 
channels of the neighboring glaciated coast. 

These features would have an important bearing on 
the question of low base-level in the district of the inside 
passages if we could be sure that the history of the 
Columbia estuary was really pre-Pleistocene; but there is 
reason to suspect that the Columbia has somewhat re- 
cently come into possession of the lower part of its valley. 
After passing the Cascade Mountains it turns northward 
in the great structural valley which farther south contains 
Willamette River and farther north holds Puget Sound. 
Then at the mouth of the Cowlitz it again turns westward, 
and traverses a low range of mountains or hills in a some- 
what narrow passage. Close to the river these mountains 
have a height of 1,000 feet or more. From the mouth of 
the Cowlitz northward to Puget Sound the country is 
comparatively low, and the summits are occupied by 


Pleistocene gravels. These features, while not demon- 
strative without further study, clearly suggest that the 
Columbia may formerly have followed the structural val- 
ley northward to and through Puget Sound, reaching the 
ocean by way of Fuca Strait. The occupation of the 
strait and sound by the great Pleistocene glacier would 
have compelled the river to find some different course, 
and when it had once carved a channel through the coastal 
hills the filling of its previous channel by glacial gravels 
would prevent its return to the earlier course. 

In view of the possibility that the lower course of the 
Columbia dates only from the Pleistocene, it is evident 
that the character of its estuary has no decisive bearing on 
the problem of pre-Pleistocene base-level. 

Summary, — Before the great glaciers of the Pleistocene 
began their work the district included a varied topography. 
The larger part was mountainous in the ordinary sense, 
with crests at various heights and a complicated system of 
steep-sided ridges, spurs and gorges. There were exten- 
sive remnants of a high-lifted peneplain, its plateaus mark- 
ing the areas of most resistant rock, and above these 
plateaus rose summits of the nature of monadnocks. 
There were remnants of a low peneplain — a peneplain 
which is now near sea-level — and these occupied areas 
of relatively weak rock. There was a system of river 
valleys or master lines of drainage, narrow where the 
rocks were most resistant and more open among weak 
rocks. The bottoms of these valleys were in part below 


Rounding of Angles, — Turning now to the results of 
Pleistocene ice erosion, one of the most evident is the 
rounding of salient features. Where the slopes of moun- 
tain spurs have an average inclination as steep or steeper 
than that which permits the resting of talus, the inter- 


mediate crest line is normally acute and serrate. The 
association of steep slopes with rounded summits is an 
abnormal condition requiring special explanation, and in 
glaciated districts there is a strong presumption that the 
rounding has resulted from the removal by ice of the 
salient parts of the spurs. The rounding, therefore, serves 
to show the extent of the district which has been subjected 
to glaciation. It also affords a rough measure of the depth 
to which the erosion has locally extended, for the imagi- 
nation restores, more or less truthfully, the original sharp- 
crested form, and thus realizes the difference between 
that and the rounded form presented to the eye. 

In the district under consideration the work of round- 
ing has been extensive. Below certain levels the crests 
and profiles of hills, mountains and mountain spurs are 
devoid of crags and sharp angles and have curved outlines. 
From this general rule there are no deviations within the 
range of our observation, except where it is evident that 
the forms of glacial sculpture have been modified by later 
work of torrents or breakers. The rounding is more 
thorough at low levels than at high. Near its upper limit 
it often amounts only to the removal of pinnacles and the 
blunting of angles which would otherwise be sharp; 
farther down it has not infrequently been carried so far 
that no suggestion remains of the pre-glacial forms. In 
many places the depth of rock pared away in the mere 
smoothing of a rough topography must have amounted to 
several hundred feet. 

The upper limit of rounding was estimated to range 
from 3,000 to 5,000 feet in the region of the inside pas- 
sage, and these estimates are in substantial agreement 
with data given by the maps and photographs of the 
Canadian Boundary Commission. From the latter I esti- 
mate the height at 4,500 feet near Behm Canal (fig. 6^) 
4,000 feet near Berner Bay (fig. 62), 5,000 feet above 


Chilkoot Lake (fig. 77), and 3,500 to 4,000 feet near 
Brady Glacier (fig. 61). In the neighborhood of Sitka it 
is about 2,000 feet. 

In a general way this limit records the extreme height 
locally attained by the confluent ice of the Pleistocene. 
As the line often runs among modern neves and glaciers, 
where glacial erosion has been in progress ever since the 
maximum ice flood, there is possibility that the later 
development of cirques has, in places, carved out sharp 
blades and pinnacles from summits that had been rounded 
by the earlier flood, but this qualification is not believed 
to be important. The upward diminution of the paring of 
salient angles is such as would naturally obtain near the 
upper limit of ice action. 

My observations of the limit of rounding are not so dis- 
tributed as to give a comprehensive picture of the extent 
of the maximum ice-sheet, 
but they agree fully with 
Dawson's conclusion that 
the whole district was oc- 
cupied. Where now are 
sounds and channels the 
ice depth was probably 

from 3)^^^ ^^ 6,000 feet, pjQ 5y^ ^ PjQRj) QF THE INSIDE PASSAGE. 

but many summits were The tops of distant mountains were smoothed 
^ ^ ^ ^, and blunted by overriding ice. 

ice-free. Some of the un- 
covered peaks were nunataks, about which ice currents 
parted to unite again. Others were the culminating 
points of mountain masses which served as centers of 

Cirques. — Other features due to ice sculpture are cir- 
ques. These occur at all altitudes above i,ooo feet, being 
most abundant in the higher regions. Those above 3,000 
to 3,500 feet now contain neves, and it is probable that they 
have been occupied not only since the last maximum of 



glaciation, but in interglacial epochs and during a part of 
pre-glacial time. Therefore, only a part of the erosion 
they represent can be ascribed to Pleistocene ice. Those 
at lower levels were made under Pleistocene conditions 
and belong strictly to that epoch. It is noteworthy that 
they occur considerably below the upper limit of ice 
sculpture; on Kupreanof Island and in Glacier Bay they 
were seen on the flanks of mountains whose summits are 
well rounded. Though these lower-lying examples are 
less fully developed than those about the higher summits, 
they represent a notable amount of ice work, and that ice 
work was performed during stages of glacier development 
intermediate in extent between the modern and the maxi- 

Fiords and Hanging Valleys, — The fiords admit of 
a partial classification as longitudinal, or strike, and trans- 


Ice-rounded mountains in the distance. 

verse. Where the courses are direct and accord with 
the general trend of the coast, and especially where two 
or more fiords or channels are parallel, it is fair to assume 
that their positions were determined by structural factors. 
Where the courses make wide angles with the trend of 



the coast, and are in detail characterized by short turns, 
it may be assumed that they are independent of strike. 
In the case of strike fiords, erosion may have been favored 
by the presence of weak rocks, but the erosion of trans- 
verse fiords had no such aid. Our best opportunities for 
direct observation were of fiords that either probably or 
possibly follow the strike, and the discrimination of 
aqueous and glacial erosion, or the problem of the amount 
of glacial excavation, is thus complicated by a factor in- 
volving much uncertainty. 

Discovery Passage and Johnstone Strait, separating 

Photographed from Johnstone Strait. In the distance, an ice-rounded summit. 

part of Vancouver Island from various minor islands and 
the mainland, constitute a well recognized fiord for 70 
miles. With little exception, its walls of rock are steep 
at the water's edge and for some distance above and be- 
low. The central depth of water ranges from less than 


200 to more than i,ooo feet. The immediate walls are 
1,000 to 2,000 feet high, curving back to rounded sum- 
mits. On the mainland side the fiord is joined by many 
troughs of similar depth and character, and by a few 
hanging valleys. From Vancouver Island it is joined by 
many hanging valleys. The sills of the hanging valleys 
best seen lie from 200 to 500 feet above tide and are evi- 
dently carved from the rock. Below each sill the con- 
tours of the main trough are continuous, without any de- 
flection toward the side valley, and the draining stream has 
only begun the work of grading its channel. A shallow 
trench is cut on the edge of the sill, and escaping from 
this, the water tumbles down the open face of the fiord 
wall. Other valleys hang so high that from our low point 
of view we could not look into them. At a moderate 
estimate the highest seen are 1,000 feet above the water, 
and as these occur opposite the deeper part of the chan- 
nel it is probable that the maximum discordance of valley 
floors is not less than 2,000 feet. All the hanging valleys 
appeared to be steep-sided glacial troughs, and those we 
saw best are at least several miles in length, with moun- 
tains behind them. 

For this complicated system of troughs I have not been 
able to suggest an origin that does not involve an immense 
amount of excavation by ice. The hypothesis demanding 
least of the ice is one which assumes the main fiord to 
follow a belt of weak rock, in which pre-glacial streams 
had sunk their beds rapidly, outstripping such small trib- 
utaries as had strong rocks to contend with. Under 
such conditions, all pre-glacial valleys, with the possible 
exception of those in the weak rock, would have been 
narrow gorges, and the work of the ice in enlarging them 
to existing dimensions would be at least as great as the 
preceding work of the streams. While this work was be- 
ing done by the tributary glaciers, the trunk glacier may 



readily have carved from the weaker rock all that part of 
the fiord trough lying below tide-level. 

The hanging valleys seen on Vancouver Island were 
evidently shaped by ice currents originating on the island 
and directed toward the mainland. During their existence 
the island contained a center of ice distribution. The 
general condition of the district at this time has been 
worked out by Dawson from studies of the striae and 
other features of ice sculpture, made along the coasts of 
Vancouver Island, of the neighboring mainland, and of 
various smaller islands in the intervening sound. The 
principal flow of ice was from the mountains of the main- 
land, taking the form either of wide individual streams or 
of a great confluent sheet; and this flood, banking against 
Vancouver Island, was divided and deflected. One great 
division, the ^ Queen-Charlotte-Sound Glacier,' moved 
northwestward to the ocean, spreading over the north end 
of the island. 
The other great 
division, the 
gia Glacier,' 
moved south- 
eastward and 
then turned 
westward to 
the Strait of 
Fuca.^ It is 
probable, also, 
that a branch of 
this stream, reinforced by tributaries from the Cascade 
Range in northern Washington, flowed southward as the 
Puget Sound Glacier. 

1 Additional observations on the Superficial Geology of British Columbia and 
adjacent regions. By George M. Dawson. Quart. Jour. Geol. Soc. London, vol, 
XXXVII, p. 278, 1881. 

FIG. 70. 




The thorough rounding of crests on Vancouver Island 
extends so far above the floors of the observed hanging 
valleys as to indicate that their glaciers were not features 
of the stage of maximum glaciation. It is quite possible 
that the greatest flood from the mainland turned back the 
feebler streams originating on the island and sent currents, 
here and there, through mountain passes to the southwest- 
ern coast. 

Along the narrow passages separating Princess Royal 
and Pitt islands from the mainland, hanging valleys are 

equally abun- 
dant, and the 
illustrations of 
the physio- 
graphic type 
are even more 
striking. The 
greater tribu- 
tary valleys 
approach the 

fiords from the mainland and have sills near water-level, 
some of them lying so low as to contain shallow bays. 
All of the sills on the side of the islands are above tide, 
and the valleys back of them are surprisingly broad when 
considered as the channels of glaciers originating on is- 
lands only fifteen to twenty miles wide. The change of 
grade from the floor of the hanging valley to the side 
wall of the main valley is so abrupt as to give the impres- 
sion that the sill is really a parapet, and that the alcove in 
the fiord wall contains a basin. At one point (fig. 70) 
we climbed to a sill with the half expectation of discover- 
ing a lake beyond, but found only the uneven, and in 
places marshy, floor of an ordinary U-trough. The high- 
est sills seen are about 1,000 feet above tide. 

As Grenville Channel, the passage separating Pitt Is- 


Ice-rounded summits in the distance. 




land from the mainland, is straight, and as it has a parallel 
on the opposite side of the island, it is confidently classed 
with the strike valleys. At each end it connects with 
much wider fiords of the mainland, having the crooked 
courses of transverse valleys. From these relations I 
infer that the pre-glacial master drainage was transverse, 
and that the Grenville valley was not followed from end 
to end by a 
stream, but 
contained two 
streams, flow- 
ing in opposite 
ways from a 
medial sum- 
mit. The pas- 
sage about 

Princess Royal Island is not clearly marked as a strike 
valley, but its narrowness as compared with the trans- 
verse valleys in which it ends indicates that it origi- 
nally contained two minor streams, with a summit be- 
tween. No trace of either summit has survived the 
glacial remodeling. As in other fiords, the bottoms are 
irregular, but some of the lowest points lie midway be- 
tween the ends. In fact, the deepest soundings reported 
in the two passages (severally 800 and 900 feet) are near 
their middles, and are not exceeded by recorded sound- 
ings in neighboring waters. When it is considered that 
these fiords, being parallel to the coast, run athwart the 
general movement of the ice from land to sea, the fact 
that their depth is comparable with that of troughs lying 
in the direction of general movement is certainly remark- 
able. It probably depends in part on the presence of a 
belt of easily eroded rock, but after all allowance for such 
favorable condition, one is impressed by the ability of ice 



to cut down its bed far below the profile which limits the 
action of running water. 

The essence of the explanation is contained in Gan- 

nett's theorem 
that the glacier- 
made valley is 
homologous, not 
with the river- 
made valley, but 
with the chan- 
nel made by the 
river. The bot- 
tom of a river 
channel is not 
evenly graded 
like its flood 
plain, but it 
abounds in hol- 
lows and hills, 
and the bottom 
of a glacier chan- 
nel has irregu- 
larities that are 
similar but on a 
larger scale. 

A phase of the 
Pleistocene con- 
dition of these 
passages is illus- 
trated along the 
base of the Fair- 
weather Range 
from Icy Cape to Cape Fairweather. A foothill ridge, 35 
miles long, is separated from the range by a nearly con- 
tinuous groove. A dozen alpine glaciers descend to the 


The inlet occupies a glacial trough entering Grenville Channel 
from the east. 




groove and are there gathered into trunks which follow 
the groove lengthwise till escape is found through some 
break in the ridge. The surfaces of trunks and tributaries 
are mutually adjusted where they join, but it is easy to 
imagine that the progressive erosion of their beds is re- 
lated to the volumes of the glaciers, and that the greater 
ice streams have the deeper channels. It is interesting 
to note in this connection that the bottom of this trough 
is below present tide-level — at least in part — in spite of 
the fact that there has here been a post-Pleistocene eleva- 
tion of land. Figure 74 shows a portion of the trough 
where it is occupied partly by glaciers and partly by an 
arm of the sea. The observer stands on the foothill ridge. 
Lituya Bay has the form of a letter T, the cross-bar ap- 
pearing in the view, and the shaft running through the foot- 
hills at the right. The large glaciers in the distance and 
foreground reach the longitudinal groove from deep 
mountain gorges. A small glacier, whose end barely 
touches the water, cascades over a sill that may be 800 
feet above tide. A hanging glacier at 2,000 feet or more 
sends a tongue down a shallow groove in the steep wall 
of the fiord. And, at the extreme left, a pocket glacier 
occupies a hanging valley at an altitude of 3,000 feet. 

The depth of the glacial excavation in the passages by 
Pitt and Princess Royal islands was probably about as 
great as in the passage by Vancouver Island. The scale 
of the topography is not far different, the evidence from 
hanging valleys is equally cogent, and the case is strength- 
ened by the presumption in favor of pre-glacial summit 
levels. Even with a low base-level it is not at all prob- 
able that pre-Pleistocene streams carried the erosion of 
any considerable part of these troughs below present sea- 
level, and the water partings within them may well have 
been a thousand feet higher. 

One of the more important passages of the Alexander 


Archipelago is so direct in its course that a straight line 
190 miles in length could be laid out upon it without 
touching either shore. It heads in the mainland at the 
extreme north and, trending a little east of south, termi- 
nates in the Pacific coast. The southern three-fifths, which 
bears the name Chatham Strait, has an average width of 
about seven miles; the northern two-fifths, known as 
Lynn Canal, averages five miles. Our direct observation 
was restricted to the northern part. 

This straightest of all the passages is also deepest. No 
soundings have yet been charted for the southern third, 
but those of the northern part indicate that a continuous 
channel can be traced with 700 feet as its minimum 
depth, and the maximum depth, as already mentioned, is 
2,900 feet. The bounding mountains, so far as we saw 
them, are 4,000 to 5,000 feet high, and the full depth of 
the trough is in the neighborhood of 6,000 feet. At its 
head the trough divides into four parts, which penetrate 
the upland, first as fiords or inlets and then as river val- 
leys. These parts diverge from their point of junction 
like the ribs of a fan, their directions ranging from north 
to northwest; and their courses are remarkably straight, 
especially in the lower parts. 

The unusual straightness of this great trough naturally 
suggests that its course was determined by some struc- 
tural feature, such as a fault or the outcrop of an easily 
eroded rock. Its breadth is in better accord with the 
second of these tentative explanations; but the matter is 
not free from doubt. If the trough is a strike valley, we 
should naturally expect to find parallel valleys associated 
with it, but the number of such is limited. A short 
parallel trough lies fifteen miles west of Lynn Canal 
and contains Excursion Inlet. East of Chatham Strait 
are Seymour Canal and Stevens Passage, which are 
approximately parallel. But a number of other features 




^ 1 


V p. 

§ £ *** .a 
* g >, g 

en O o 

" :« ^ •- 2 


,M 08 \. o 


5 a « « 

^y^ .^ "^ ft *** 

C « " «" 

C «, <u ^. 

an o-g 

^ <fl rt iS 

K >»«« jy 

^ ^ o ^ 

-^5 8 

•a -2 2 d 

ftg is c S 
Qj rt H ce «i; 

t; ij 2 u fl 

ft' fT ?> «i ** 
^ .H .^ 5 2 
aw H cd • 

5 -2 ^ ^ ° 

H t 

cd O 



in close association exhibit a systematic northwest trend; 
these are Icy Strait, Freshwater Bay, the main part of 
Tenakee Inlet, and Peril Strait with its prolongation in 
Hoonah Sound. The divergence of the branching fiords 
at the head of Lynn Canal is also perplexing; for, while 
each of these is so straight as to suggest an origin con- 
nected with strike, they are not parallel, but diverge fan- 
wise from a common point. 

The walls of Lynn Canal are well-defined features. At 
the water's edge and for some distance above, their con- 
tours are simple. There are few bays and no jutting prom- 
ontories. The slope of the walls is not so steep as in 
some fiords; it rarely exceeds 45°, and in places is as low 
as 30°. Near the head of the canal the walls are well 
defined up to an altitude of about 2,000 feet, the height 
increasing toward the north and decreasing toward the 
south. Higher up, the mountains exhibit a varied topog- 
raphy: There 

are U-shaped 
troughs end- 
ing as hang- 
ing valleys at 
many differ- 
ent heights; 
there are V- 
shaped gorges 
modified at 
bottom by 
glaciers and 
separated by 
narrow tenti- 

form spurs. These troughs, gorges and spurs approach 
the canal wall at all angles and there end. Some- 
times they end abruptly, sometimes with a certain amount 
of prolongation down the steep wall; but this prolon- 


The sill of the hanging valley is 3,500 feet above sea-level. For 
more than 1,000 feet below the glacier the fiord wall is kept bare by 
falling fragments of ice. 



gation never extends to the water, and usually does not 
affect the slope within one or two thousand feet of the 
water. Their relation is such as might arise if the topog- 
raphy of the upland had formerly extended farther in the 
direction of the main trough and afterward been truncated 
by the development of the trough. 

The branch troughs which unite to form the canal 
have similar walls, which rise higher before meeting the 
varied topography of the upland. They are also narrower, 
and at least two of them contract in their upper parts so 


Below is Chilkoot Lake, and beyond are Chilkoot Inlet and I<ynn Canal. The lower part 
of the mountain was shaped by a great Pleistocene glacier, the upper by small tributaries, 
which survive. The summits are ice-rounded nearly or quite to 5,000 feet. The floors of hang- 
ing valleys are at 3,500 to 4,000 feet. 

that the cross-profile is more nearly a V than a U. Skag- 
way Canyon, a tributary to the Taiya trough, is narrow at 
bottom, except where occupied by alluvium. Glaciation 
has smoothed its walls on a grand scale, and has degraded 
its bottom enough to render the floor of a tributary dis- 
cordant to the extent of 50 or 100 feet, but the type of 
cross-section acquired from pre-glacial stream erosion has 
not been destroyed. Photographs show that the upper 



part of Taiya Valley has a similar character, and it 
seems to me probable that these gorges have been only 
moderately deepened by glaciers. The pre-glacial stream 
grades which they suggest would pass below present tide- 
level before reaching the head of Lynn Canal, and would 
be adjusted to a quite low base-level at the south end of 
Chatham Strait. 

The bottom of the main trough is characterized by much 


The profile follows the lowest line through Taiya Valley, Taiya Inlet, Lynn Canal and 
part of Chatham Strait, Whole distance 165 miles. Vertical scale about ten times the hori- 
zontal. Base line 3,000 feet below sea-level. The approximate height of mountains east of 
the trough is indicated. 

irregularity. The line of deepest water is sinuous and 
often wanders far from the middle, and along this line the 
depth varies irregularly (fig. 78). The soundings are 


Based on contour map by Canadian Boundary Commission and soundings by U. S. Coast 
Survey. The positions are indicated by corresponding letters in figfure 78. Profile A crosses 
Chilkoot Lake (fig. 75), Ferebee Inlet and Taiya Inlet. Profile J? crosses Chilkat Inlet, near 
Davidson Glacier, and Lynn Canal. Profile C crosses Lynn Canal below Sullivan Island. 
S. L., sea-level. Base line 3,000 feet below sea-level. Vertical scale same as horizontal. 

too far apart to give expression to details of configuration, 
and we do not know to what extent the irregularities may 


be ascribed to unequal deposition of drift, but there is 
reason to believe that much of the inequality pertains to 
the rock floor and is to be ascribed to glacial erosion 
as conditioned by varying resistance of the terrane. A 
number of islands and one long peninsula are so placed 
that they may properly be regarded as portions of the 
trough bottom which rise too high to be covered by the 
sea. They are of rock, are thoroughly glaciated, and their 
axes, as well as all flutings and other lines of sculpture, are 
parallel to the fiord walls. Ice erosion has gone so far as 
to destroy all semblance to the forms characteristic of 
aqueous sculpture. Several islands and the southern 
part of the peninsula are shown in figure 3, and the view 
in figure 75 was made from the summit of the penin- 
sula, 1,750 feet above sea-level. 

From these various facts, as well as from general im- 
pressions to which it is not easy to give definite expres- 
sion, I would draw the following tentative inferences : The 
V-gorges at the heads of branches of Lynn Canal and in 
the uplands bordering the canal, were made chiefly by 
pre-glacial streams, and they have been but moderately 
deepened by glaciers. There was a pre-glacial, com- 
paratively narrow, valley through Lynn Canal, the floor 
of the valley being below present sea-level and related to 
a low base-level. Lateral V-gorges were tributary to 
this and were largely adjusted to it in grade. The Pleis- 
tocene glacier broadened the river valley, truncated the 
side spurs and the tributary gorges, and at the same time 
materially deepened the valley for the whole breadth of 
the trough. The glacial degradation is conceived as 
averaging hundreds of feet and possibly more than one 

The lateral valleys so situated as to carry glaciers under 
other than Pleistocene conditions, are thoroughly shaped 
in characteristic U forms, and they now contain glaciers. 


There are also valleys which are now free, or nearly free, 
from ice, and which were filled with ice only in the 
presence of the great glacier, and these have less charac- 
teristic glacial forms. It would seem that the presence 
of the great glacier in Lynn Canal gave them so high a 
base-level that their ice streams were sluggish and had 
little power of erosion. 

The valley of Davidson Glacier deserves special men- 
tion as an illustration of the broad contrast which may 
exist between the systems of currents associated with 
different stages of glaciation in the same district. The 
neves supplying the glacier are not in sight from Lynn 
Canal, but lie back of the first line of summits, and the ice 
river flows eastward through a lateral or transverse valley, 
entering the great fiord approximately at right angles. 
During the great ice flood the general direction of move- 
ment was toward the south, and Davidson Glacier did not 
exist; its valley was not only filled with ice but over- 
ridden. It results that the upland bordering the Davidson 
trough in the vicinity of the Lynn trough, is planed and 
fluted with lines trending southward, while the Davidson 
trough, being shaped wholly by the Davidson Glacier dur- 
ing epochs of moderate ice supply, has an entirely inde- 
pendent sculpture, the lines of which are transverse to 
those of the overlooking upland. The Davidson trough 
is a fine example of its type. So far as visible from the 
sea, it is of uniform width; its parallel walls sweep in a 
simple curve of large radius and are steep. Although 
the topography just above and back of them is varied, they 
themselves have neither salient nor reentrant angles. The 
glacier has evidently adjusted its channel quite com- 
pletely to the conditions of its flow, and has at the same 
time sunk itself deeply into the mountain it traverses 
(see fig. 2). 

The transverse troughs of the mainland we did not 



visit; they are known to me only through photographs, 
the contour maps of the Canadian Boundary Commission, 
and the soundings of the United States Coast Survey. 
As already mentioned, they are judged to be transverse 
to the strike, because they make wide angles with the 
general trend of the coast and because they are charac- 
terized by many short 
turns (fig. 80). Be- 
ing independent of 
strike, their courses 
are also independent 
of variations in rock 
texture, and this 
character makes 
them specially avail- 
able for the study of 
glacial erosion. 
Their walls are steep; 
photographs give the 
impression that they 
are decidedly steeper 
than those of Lynn 
Canal and its 
branches, but this 


The river is rapidly filling its deep trough with alluvium. ciV)lv duC tO thc tcnd- 
Glacial rounding extends above the base of clouds, which ^ 

cut oflFthe view at about 3,000 feet. Cncy of photOgraph- 

ers to select localities exhibiting bold scenery. The 
walls stand well apart, after the habit of glacial troughs, 
and exhibit notable parallelism (fig. 62). Hanging valleys 
abound, ranging in height up to 3,000 feet or more and being 
more numerous at the greater altitudes. Some, like the 
Stikine and Whiting, are occupied by rivers ; others, like 
the Taku, Speel, Unuk and Skeena, contain rivers in their 
upper parts and admit the sea below; others, like Tracy 


Arm, Behm Canal and Portland Canal, have the character 
of fiords through their whole extent. Where rivers flow 
through them, the whole width from wall to wall is 
occupied by alluvium, and a lace-work of channels indi- 
cates rapid deposition. The open water has the depth, 
and the irregularity of depth, characteristic of fiords. In 
Tracy Arm, which has a breadth of one mile and a length 
of twenty, the soundings range from 850 to 1,150 feet. 
The main part of Behm Canal, with a width of two to 
three miles, has a depth ranging from 1,100 to 1,850 feet. 
In Portland Canal, which for eighty miles has an average 
width of two miles, the range of soundings is from 300 to 
2,100 feet, and a depth of 1,200 feet occurs near the head. 

It is scarcely to be doubted that all these long troughs 
were initiated by pre-glacial rivers; but the rivers could 
not have opened their valleys to the present width with- 
out giving time for the breaking down of the walls. If 
the river valleys were deep, they were also narrow, and 
their widening was the work of the Pleistocene glaciers. 
In the work of widening, the more obstructive projections 
were removed and the contours of the walls were simpli- 
fied. The amount of erosion necessary to convert the 
assumed V-gorges into the observed U-troughs is large, 
and the ice streams by which it was done could not have 
failed to wear down the floors of their channels at the 
same time. While it is quite possible that the down- 
stream parts of the river gorges had been sunk below 
present sea-level, the greater part of the excavation below 
sea-level was probably performed by the glaciers. 

Inequality of Glacial Erosion, — The great work 
which it has seemed reasonable to ascribe to ice in the 
deepening and widening of fiords and other troughs stands 
in striking contrast to the feebleness of ice erosion in 
other places, which permitted, for example, the preserva- 
tion of the low peneplains of Annette Island and the 


vicinity of Sitka. In the one case the depth of the ero- 
sion is measured by hundreds of feet, in the other by 
tens. To a certain extent inequalities of erosion were 
determined by inequalities of resistance, but as the rock 
of the low peneplain is not of notably resistant character, 
and the rock traversed by the transverse troughs is in part 
of highly resistant character, it is evident that this is 
not the dominant factor. To an important extent also 
differences of erosion were determined by differences in 
the depth and consequent pressure of the flowing ice. 
But I conceive that the most important of the variable 
factors was the velocity of the ice currents. At the height 
of the Pleistocene flood the snow-fields were on high pla- 
teaus and mountain masses, from which the ice crept in 
broad sluggish streams to the preexisting channels of 
drainage. In these channels it assumed the character of 
rivers, and the lines of pre-glacial water drainage became, 
in the main, the lines of Pleistocene glacial drainage. 
Along the deeper waterways the ice could flow more 
rapidly because its depth was greater, and its ability to 
erode was correspondingly increased. Thus it was that 
the old river gorges, being adopted as lines of flow by the 
ice, were widened, straightened and deepened, while 
the adjacent uplands received comparatively little modifi- 
cation. The remnants of low peneplain were preserved, 
despite the softness of their rocks, because they lay out- 
side the lines of flow of the strong currents. The remark- 
able deepening of the fiords of the mainland is probably 
connected in part also with the fact that they were longer 
occupied by glaciers than were the channels of the archi- 
pelago. Many of them still have glaciers at their heads ; 
others are flanked by glaciers which would descend and 
fill them should the climate swing but slightly toward the 
Pleistocene condition. 

The more thorough rounding of salient angles at low 


levels than at high is probably due to the concurrence of 
several causes rather than to a single one. First may be 
mentioned the difference in pressure. The rate of abra- 
sion, and of other forms of erosion, is probably enhanced 
by pressure, and during the height of the glacial flood all 
low-lying surfaces were subjected to greater pressure than 
those at higher levels. A second cause is connected with 
duration. The coming on and the passing of each Pleisto- 
cene epoch of glaciation was probably gradual, and there 
were doubtless glacial epochs of less intensity than the 
maximum. Thus the crests of low hills were subjected 
to glacial wear for longer periods than those of high spurs, 
and the difference may have been very great. Some al- 
lowance may also be made for subsequent modification at 
high levels. Ever since the last great ice flood began to 
wane, the valleys between high spurs have been occupied 
by small glaciers, and these have tended, by the develop- 
ment of their valleys, to reduce the width of the separat- 
ing ridges. Thus, while the rounding at low levels was 
still in progress, a work had been begun which tended to 
reduce the effect of the rounding at high levels. 

Glacial Deposits, — In the narrower parts of the inside 
passages we saw no accumulation of glacial drift. It is 
possible that drift masses were concealed here and there 
by the dense forest, but usually the spaces between cliffs 
and other visible outcroppings of rock were not large. It 
is probable that drift masses are concealed by the water, 
and have share in the production of the irregularity of the 
bottom; and a few bars and dams which are known to 
interrupt the continuity of other fiords are presumably 
morainic. As it is natural that the routes of travel should 
avoid such obstructions, it would be rash to infer the ab- 
sence of local moraines from our failure to observe them. 
The only important bodies of Pleistocene drift which we 
saw are on the shores of the Gulf of Georgia, where 


several extensive banks of waterlaid material constitute 
terraces, both on the mainland and on the shore of Van- 
couver Island. These banks are flat-topped, and esti- 
mated to rise from loo to 200 feet above the water. One 
of them was seen to overlap hills with distinctly glacial 
sculpture, and as they are from 50 to 100 miles within 
the limit of the glaciated district they must have been 
accumulated after the last ice maximum. 

The glacial deposits we encountered are of trivial mag- 
nitude collectively, in comparison with the glacial erosion 
of which we saw evidence, and it was therefore inferred 
that the principal regions of glacial deposition lay outside 
the field of our observation. This inference agrees with 
the conclusion of Dawson that the ice-sheet embraced the 
entire Alexander Archipelago, together with all other is- 
lands of the coast except the Queen Charlotte, and that its 
outer margin was beyond the present line of coasts. 

Associated Sea-Levels. — The question of the relations 
of sea and land at the time of the great Pleistocene glacia- 
tion is of much interest, and some considerations bearing on 
the question will be mentioned, although the evidence at 
present available is either indirect or negative. As gla- 
ciers are chiefly phenomena of the land, and as glacial 
erosion in this district has been carried far below the 
present sea-level, it is natural to assume that the sea- 
level associated with that erosion was much lower. A 
little consideration, however, will show that such a con- 
clusion does not necessarily follow. The deepest known 
hollow ascribable to ice work is in Chatham Strait, and 
lies 2,900 feet below sea-level. At that point the total 
depth of the great glacier was probably 6,000 feet. It is 
commonly assumed that where a glacier enters a sea not 
deep enough to float it, a part of the ice, equal in weight to 
the displaced water, is upheld by the water, and the pres- 
sure of the glacier on its bed is correspondingly dimin- 


ished. While I do not regard this assumption as valid,^ 
I entertain it for the moment because it gives a minimum 
estimate of the pressure of the Chatham Strait glacier. 
If the then sea plane had the same height as the present, 
the pressure of the glacier would be modified, by assump- 
tion, by the sustaining power of 2,900 feet of sea water. 
This sustaining power is equivalent to about 3,300 feet 
of the total ice thickness, leaving 2,700 feet of ice to press 
upon the bottom of the strait; and such a pressure would 
manifestly be ample for the work of erosion. So far as 
our numerical data go, this locality affords an extreme 
case; and as the hypothesis of glacial erosion below 
present sea-level is not barred for this locality, it is prob- 
ably not barred for the whole district. 

Another consideration is connected with the reaction 
of the ocean on the front of the glacier. My own obser- 
vations, though comparatively limited, are so accordant 
with Dawson's generalization that I accept with confi- 
dence his conclusion that the Pleistocene ice front lay 
outside the present coast line throughout practically the 
whole district. If the ocean had then its present level, it 
washed the ice front for hundreds of miles. The power 
of the ocean to waste a glacier by melting is ordinarily 
greater than the power of atmospheric agents, and along 
the present coast of Alaska is much greater. As the fac- 
tors are complex, it would be difficult to give an analytic 
demonstration of this proposition, but it is easy to illus- 
trate by an example. The great confluent glacier which 
filled Glacier Bay in the eighteenth century had a general 
depth of 1,000 to 2,000 feet along the axis of the bay, and 
rested with less depth on adjacent tracts of land. All 
through the nineteenth century it was depleted, its wast- 
ing being brought about partly by the sea and partly by 
atmospheric agencies. The sea not only melted back the 

1 See discussion in chapter ni. 


front for 35 miles, but disposed of a continuous accession 
of ice from the Muir, Grand Pacific, and other glaciers. 
Of that portion of the confluent glacier which lay upon 
the land, a part slid into the sea and was there melted, but 
another part, which happened to rest on comparatively 
flat surfaces, remained on the land and has been subjected 
to only atmospheric agencies of waste. These agencies 
have not yet completed its destruction, so that extensive 
patches of stagnant ice, receiving no accessions from 
neves, remain to testify to the comparative feebleness of 
the atmospheric attack. I think there can be no exag- 
geration in the estimate that the melting by the sea has 
exceeded by ten times the melting from direct insolation 
and the contact of warm air and warm rain. It is further 
to be noted that this work of the sea was performed in a 
land-locked bay, where water cooled by the ice tends to 
accumulate and is but slowly exchanged for the warmer 
water of the open ocean. On the outer coast, where the 
supply of warm water would be constantly renewed by 
currents, the melting power would be still greater.^ 

The warmth of the Pacific Ocean in this region depends 
chiefly on the great circling current of the North Pacific, 
a part of the planetary system of oceanic circulation which 
might be modified but would not be stopped by any causes 
we may suppose to have existed in Pleistocene time. All 
through the Pleistocene the melting power of the ocean 
along this coast must have been great, and it is not prob- 
able that glaciers could so far have withstood it as to ad- 
vance in deep water. The features of another part of the 
Alaska coast, to be presently described, render it probable 
that the glaciers were checked by the oceanic melting 
somewhere in the belt of shallow water, and were com- 

1 Muir Inlet near the glacier is 7,cxx) feet broad and 500 feet deep. Here Reid 
found a general temperature, at all depths, of 38° F. The oceanic temperature 
in the Gulf of Alaska is 58° F. The water of the inlet is 6 degrees warmer than 
melting ice, that of the neighboring ocean 26 degrees. 


pelled there to yield up their load of rock waste. We 
know, from the extent of Pleistocene erosion in this dis- 
trict, that the burden of rock waste was large, and wher- 
ever the moraines were built they made an important de- 
posit. If the Pleistocene base-level had been the same as 
the present I should expect to find that deposit as a con- 
tinuous bar, or string of linear islands, along the outer 
coast of the Alexander Archipelago and at various points 
farther south, but no such features have been described, 
and the coast seems to have an entirely different character. 
This consideration, though connected at present with only 
negative evidence, distinctly favors the theory that the sea- 
level associated with the greatest Pleistocene ice floods 
was considerably below the modern sea surface. 

There is some evidence, on the other hand, of a com- 
paratively high sea plane after the glacial maximum. 
Dawson describes an extensive marine deposit, reaching 
a height of 200 feet, on the east side of Graham Island, the 
most northerly of the Queen Charlotte group, and infers 
from its relations that it was contemporaneous with a 
development of local glaciers.^ At Nanaimo, on the 
inner coast of Vancouver Island, he found shell-bearing 
marine clays, resting on glaciated rocks, at a height of 70 
feet above the sea.^ In Gastineau Channel is a narrow 
shore terrace at a height of 200 feet, and in an associated 
clay Dall found marine shells. At two, at least, of the 
localities the shells are of species indicating cold water, 
and each locality is on the border of a sound or channel 
which would be filled with icebergs by a moderate devel- 
opment of glaciers. If the phenomena all belong to the 
same chapter of Pleistocene history, they record an 
episode similar to that of the Champlain clays of the At- 
lantic seaboard. 

* Quart. Jour. Geol. Soc. London, vol. xxxvii, p. 281, i88i. 
« Quart. Jour. Geol. Soc. London, vol. xxxvii, p. 279, 1881. 


The terrace about Gastineau Channel is rather con- 
spicuous at Douglas, where the forest has been removed, 
and was detected at several points where the forest still 
stands, but I searched in vain for its continuation about 
the shores of Lynn Canal and Cross Sound. 


Between Cross Sound and Prince William Sound are 
high mountains visible from the sea. They are much 
loftier than those of the Inland Passage district, and 
probably comprise a number of distinguishable ranges, 
though their system is not yet known. A mass cul- 
minating in Mount St. Elias (i8,ioo feet), west of Yak- 
utat Bay, has been called the St. Elias Alps, and this 
name is sometimes made to include the whole chain. A 
more easterly portion, culminating in Mount Fairweather 
(15,000 feet), is sometimes distinguished as the Fair- 
weather Range. Opposite these mountains the coast line 
is comparatively simple, being interrupted by fiords at a 
few points only and having no important islands. From 
Icy Cape westward the mountain base is bordered by a 
low foreland, narrow at first, but broadening to ten miles 
at Alsek River and fifteen miles at Yakutat Bay and 

Many alpine glaciers now creep down the face of the 
range and unite in plateau masses on the foreland, a 
few of them spreading to the sea. The foreland bears 
also a system of ridges, of peculiar type, which are be- 
lieved to be morainic and of Pleistocene age. Mention 
has already been made of one of these in describing the 
La Perouse Glacier. The ridge against which that glacier 
crowds, at the point of our visit, runs parallel to the coast 
for about seven miles and has an actual height of 1,000 
feet, as indicated on the map of the Canadian Boundary 
Commission. It is nearly straight, is steeper toward the 


land than toward the sea, and is contoured on the sea- 
ward side by a terrace, which probably lies 200 to 300 
feet above tide. At the single point of examination this 
terrace contains marine beds overlain by glacial gravel, 
and the ridge back of it appears to consist of unsorted 
drift. The clays, where seen in section, are in part level 
and in part disturbed (page 41). The completion of 
the moraine ridge was subsequent to the deposition of 
the clays. The undisturbed clays imply, by their posi- 
tion, a higher stage of the ocean, and the overlying 
gravels are more readily explained by assuming that the 
sea-level was still high at the time of their deposition. It 
seems to me probable that the moraine ridge was depos- 
ited when the sea stood several hundred feet higher than 
now against the land. 

Toward the southeast the ridge ends against the side 
of the La Perouse Glacier, and the relations indicate that 
the glacier is wearing it away . At one time it probably 
extended considerably farther in that direction, but was so 
low as to be overridden by the modern glacier and thus 
subjected to erosion by it. 

Lituya Bay, a few miles up the coast toward the north- 
west, penetrates the land for seven miles, dividing not only 
the foreland but a low outer range of mountains. Within 
the mountains its trough has fiord characters, and from the 
walls of the mountain gateway run two morainic ridges, 
parallel at first, but curving toward each other and de- 
scending so as to unite under water at the entrance to the 
bay. These were seen only from the ship's deck, but 
their character is unmistakable. They are steeper within 
than without, and their inner contours continue those of the 
passage through the mountain ridge. They extend some- 
what beyond the general line of the foreland, making a 
pair of capes which embrace the outer part of the bay. 
Close to the mountains they have an extreme height of 


i,ooo feet above the sea, as indicated by the contours of 
the Canadian Boundary Commission's map. 

Between Lituya Bay and the mouth of Alsek River 
are several other morainic ridges, more or less crescentic 
in form, and so related to modern glaciers as to indicate 
that they were formed by them at some earlier epoch, when 
the ice streams were greater than now. In some cases 
the modern ice seems to extend to the base of the mo- 
rainic rampart, and in one instance ice projects through a 
gap in the rampart; but the height of the old moraines 
forbids the idea that they were adjusted to glacial condi- 
tions closely resembling the present. From a distant 
view some were judged to be as high as 1,500 feet, 
and the modern glaciers seemed merely to touch their 
bases, instead of pressing against their inner faces. Asso- 
ciated with them, and extending along the coast beyond 
them, are a system of terraces, the highest of which was 
estimated to be 500 feet above tide; and these suggest 
that the sea surface stood comparatively high when the 
moraines were built. 

The fact that the old moraines rise far above the modern 
glaciers but yet inclose areas only a little larger than the 
modern ice is able to occupy, is notable, and doubtless has 
an important significance if rightly interpreted. The ex- 
planation which occurred to me is, that the extent of the 
old glaciers was restricted by contact with sea water. As 
already pointed out, a warm ocean, like the Pacific in the 
Gulf of Alaska, is a most efficient agent for the wasting 
of glaciers. The high specific heat of water, the freedom 
of the circulation by which warm water is brought to re- 
place the water which has parted with its heat, and the 
great depth of the body of warm water on which this cir- 
culation can draw, all contribute to this efficiency. It 
seems to me entirely possible that if the greater glaciers 
descending from Fairweather Range in earlier times en- 


countered the sea soon after reaching the basal plain, 
they may have been wasted so rapidly along the line of 
contact as to determine there a principal line of morainic 
accumulation, and the ramparts of the coast may constitute 
the record of a successful resistance by the sea to glacial 

If this explanation is correct, the Pleistocene submer- 
gence of the coast probably extended considerably above 
the zone of terraces. The oceanic resistance determining 
the deposition of drift would not be effective after the 
moraine ridge had been built so high as to become a 
partition between ice and water, and the theoretic posi- 
tion of the old sea plane is therefore along or above the 
highest crests of the moraine. 

The hypothetic conditions are similar to those at the 
foot of Davidson Glacier. If that glacier should shrink, 
and the sea-level should be lowered in Lynn Canal, the 
Davidson moraine, if composed wholly of rock debris, 
would survive as a crescentic rampart with a flat top; 
and if the ramparts of the Fairweather region are strictly 
homologous they should have broad and level crests. If, 
however, the Davidson moraine is not wholly built of 
rock debris, but consists rather of an apron of debris 
resting against a concealed ice slope, the resulting ram- 
part, when ice and water are withdrawn, would have an 
acute crest, with uneven sky-line ; but the crest, instead of 
representing the plane of the present water-level, would 
stand somewhat below it. In viewing the old moraines 
of the Fairweather coast I recognized a few local flat 
summits, but the general character of the crest line is 
acute and its height is not uniform. If, therefore, the re- 
lation of the old glaciers to the sea was like the present 
relation of the Davidson Glacier, the sea then stood much 
higher against the land than now — or else the land then 
stood lower. 


Whether these old ramparts represent the limit of 
Pleistocene ice at the time of maximum glaciation is a 
question as to which there may be doubt, but I incline to 
the view that the affirmative answer is the true one. So 
far as I could judge from distant views, the rounding of 
rock ridges and crests, characteristic of flooding by ice, 
extends but a moderate distance above the surfaces of 
modern glaciers on this part of the coast, and indicates 
glaciers of about the same magnitude as are indicated by 
the rampart moraines. 

On the northwest side of Dry Bay, which receives 
Alsek River, a single fragment of a high rampart was 
seen, and in association with it a terrace; but thence to 
Yakutat Bay the broad foreland is low, its flatness being 
relieved only by ridges of moderate height. About Yak- 
utat Village and Ocean Cape, at the mouth of Yakutat 
Bay, some of these ridges were seen to be morainic, 
and it is supposed that the whole foreland is constituted 
of glacial waste, chiefly of waterlaid gravel. If high ram- 
parts were formed along this portion of the coast they 
were afterward destroyed and the material carried sea- 
ward by later advances of the ice. 

On the mountains near the head of Yakutat Bay are 
fragmentary terraces at various heights ranging up to 
1,200 or 1,500 feet, but as they face the bay rather than 
the ocean, it is entirely possible that they are of glacial 
rather than marine origin. Hanging valleys occur on the 
eastern wall of Yakutat Bay and on the walls of Russell 
and Nunatak fiords. One of these, near Nunatak Fiord, 
is now occupied by a glacier which cascades for 1,000 
feet down the steep wall of the fiord. The sills of these 
valleys range from tide-level to a height of 1,500 feet or 
more, and they tell of great erosion by the trunk glaciers. 
The associated rounding of topographic angles is carried 
higher above surfaces of modern glaciers than in the 



vicinity of Mount Fairweather, and this character seems 
naturally related to the gentler declivity of the mountain 
front. The enhanced alimentation of Pleistocene glaciers 
would tend everywhere to increase their thickness and 
their rate of flow, but it is easy to understand that the 
greater resistance to flow encountered on a gentle slope 
would cause 
the thicken- 
ing there to 
be more nota- 
ble than on a 
steep slope. 

So far as 
may be judged 
by the gradi- 
ents of the 
Malas pi na 
Glacier, the 
ice flood as- 
sociated with the hanging valleys and rounded crests of 
the Yakutat region should have extended farther seaward 
than the line of the present coast, and it is probable that 
the outer morainic deposit is not visible. The submerged 
moraine ridges within the bay and across its mouth (page 
49) pertain to quite moderate expansions of the Malaspina 

The inference that the sea-level associated with the 
moraine ramparts coincided with their highest summits 
may be interpreted in terms of land change or sea change. 
The local phenomena would be explained by assuming 
that the land has risen about 1,500 feet since the building 
of the moraines, or by assuming that the sea has subsided 
that amount, and it is, of course, possible that there have 
been changes of both land and sea, and that the discord- 



ance of levels represents their sum or difference. While 
the question thus raised is not susceptible at present of 
a complete answer, it is nevertheless possible to make 
some progress in that direction. 

In studying the Alexander Archipelago, which adjoins 
the Fairweather coast on the southeast, no evidence was 
found of a high sea-level in association with the greatest 
Pleistocene glaciers; but it seemed probable, on the con- 
trary, that the mid-Pleistocene sea-level was considerably 
lower than the modern sea-level. Beyond the St. Elias 
Mountains, in the opposite direction, lies the district of 
Prince William Sound, and the Pleistocene history of that 
region appears to resemble closely that of the Alexander 
Archipelago. The Pleistocene history of the Fairweather- 
St. Elias coast thus appears to be exceptional and to be con- 
trasted with the histories of neighboring coasts on both 
sides. This contrast is associated with a contrast in gen- 
eral geologic history, as revealed in the physiography. 
The district of the Alexander Archipelago is genetically 
a plateau, from which mountains and valleys have been 
developed by erosion. The district about Prince William 
Sound has been found by Schrader and Spencer^ to have 
the same character and history. In each case a region of 
complex structure was reduced to a condition of low 
peneplain by long-continued erosion and then uplifted 
bodily though somewhat unequally. The original altitude 
of one plateau ranged from 3,000 to 7,000 feet, that of the 
other averaged 6,000 feet. The intervening tract has 
been lifted to a much greater height, so that its culminat- 
ing peaks have altitudes of 15,000 to 19,000 feet, and it is 
probable that crustal deformation here produced mountain 
ranges directly, instead of creating a plateau from which 
they were developed by erosion. 

1 Geology and Mineral Resources of a portion of the Copper River district, 
Alaska. U. S. Geol. Survey, pp. 62-76, 1901. 


Two facts indicate that these mountains are geologically 
young. The first is paleontologic. Russell found, in 
one of the lower spurs of Mount St. Elias, a fossil marine 
fauna composed wholly of forms which still inhabit the 
coastal waters of Alaska.^ These show that the last great 
elevation of the mountain range is recent, as measured in 
terms of biologic evolution. The other evidence of youth 
is found in the great height of the mountains. As pointed 
out by Powell, the degradation of mountains is so rapid 
that only young mountains can be lofty. The St. Elias 
Range is not only lofty but steep, and its rate of waste 
must be rapid. The fact that it is lofty despite rapid 
waste indicates that its waste is compensated by growth. 

In view of the differences in general geologic history, 
there need be no surprise if the Pleistocene history of the 
district of high mountains should differ from the Pleisto- 
cene history of the districts of Alexander Archipelago 
and Prince William Sound. In view of the loftiness of 
the mountains, it is rather probable than otherwise that 
uplift has occurred since the epoch of chief Pleistocene 
glaciation. It is therefore inferred with some confidence 
that the discordance between the sea-level indicated by 
the rampart moraines and the present sea-level has been 
brought about chiefly by local uplift of the land. 

If this view is correct, the disturbed marine clays ob- 
served near La Perouse Glacier (fig. 22) may be con- 
nected with a fault zone of Pleistocene or post-Pleistocene 


Prince William Sound is a very irregular bay, opening 
southward (pi. xiii). All about it are mountains, the 
higher being massed at the north, and others encroaching 
on its area as promontories and islands. The largest islands, 

1 National Geographic Mag., vol. 3, pp. 171-172, 1891. 



Montague and Hinchinbrook, lie at the south, separating 
the sound from the open ocean. The inlets of the north- 
ern coast, some of which we visited, are fiords, abounding in 
evidence of glacial sculpture, and the lower slopes of the as- 

Shows the glacial rounding of mountain crests, fifteen miles from the ocean. 

sociated peninsulas and islands are well rounded. On the 
east side of the sound the rounding extends to an estimated 
height of 3,000 feet, in Columbia Bay, at the north, to about 


Shows a serrate crest line, little, if at all, modified by overriding ice. 

4,000 feet, and to the same or greater height in Port Wells. 
Seaward the height diminishes and the higher crests of 
the outer islands are comparatively angular and serrate. 


It is probable that Pleistocene ice occupied the whole 
sound, but nothing is known of its seaward limit. 

In College Fiord, a branch of Port Wells already de- 
scribed (page 8i), there is a magnificent system of hanging 
valleys, the larger being still occupied by glaciers, which 
enter the fiord with ice cascades. These side glaciers 
have accomplished something in the way of erosion since 
the disappearance of the trunk glacier to which they were 
once tributary, so that they do not rest upon the un- 
modified face of the fiord wall, but occupy shallow chan- 
nels. In a general way the surfaces of the glaciers, in 
their lower courses, are flush with the adjoining portions 
of the fiord wall. 

My attention has been directed by Gannett to the fact 
that several of the cascading glaciers make two leaps, and 
that there is a certain amount of harmony in the spacing 
of the falls. When the region shall have been thoroughly 
studied it is possible that the interpretation of these corre- 
spondences may develop a special chapter in the history 
of the ice retreat. 

With the aid of a series of photographs made by Mer- 
riam, I have computed the approximate heights of the 
more important cascades, as follows: Wellesley, 1,700 
feet; Vassar, 2,200; Bryn Mawr, (trunk) 1,300, (left 
branch) 2,700, (right branch) 2,500; Smith, 1,250, 1,700 


S ' z 

Well Vas. B M. Smith Rad 


Short horizontal lines show the relative positions of the cascades of Wellesley. Vassar, 
Brj'n Mawr, Smith and Radcliflfe glaciers. S.L. sea-level. Scale, 14,000 feet = i inch. Com- 
pare figures 44 and ^5. 

and 2,600; Radcliffe, i,8oo and 3,500. When these are 
platted to scale in their proper vertical and horizontal re- 
lations (fig. 84) they fall into two series, descending south- 
ward from the head of the fiord. Making some allowance 


for the greater volume of the side glaciers when the trunk 
glacier filled the fiord, I have indicated the profile of the 
trunk glacier by a dotted line (ab). The inclination of 
this line from the horizontal is about 2°, or one in twenty- 
five. Its height above tide ranges from 2,800 to 4,800 feet, 
and it indicates a thickness of ice exceeding these figures 
by the depth of the fiord, whatever that may be. In the 
line of Gannett's suggestion, a second tentative profile 
(cd) is drawn in similar relation to the crests of the lower 
series of cascades. 

The depth of ice indicated by the hanging valleys is 
somewhat less than that which would be inferred from 
the rounding of projections, and it seems probable that 
the epoch during which the hanging valleys received 
their principal sculpture was not the epoch of maximum 

A cordon of high hanging valleys surrounds Harriman 
Fiord. Above Barry, Serpentine and Surprise glaciers 
they contain hanging glaciers at a general height of about 
4,000 feet, and east of Harriman Glacier their ice banks 
coalesce in a continuous terrace along the valley wall (page 
95). The surface of the trunk glacier to which they are 
adjusted probably lay 5,000 feet above present sea-level. 

On the north side of Montague Island and at various 
points on the peninsulas of the east and west sides of the 
sound, a horizontal terrace was observed, at an estimated 
height of 50 to 75 feet above tide. No near view was 
obtained, and I did not learn its character. 

As the Pleistocene glaciers extended at least to the 
outer coast line, as their work of erosion was great, and 
as their limit is not indicated by conspicuous moraines, the 
provisional inference is made, as in the discussion of the 
Alexander Archipelago, that the ocean surface was com- 
paratively low at the time of their greatest expansion and 
that their outer moraines are now submerged. 



My observation of the Kenai Peninsula was restricted 
to the northwest side of its more southerly arm. It is 
there constituted by a lofty upland partially dissected by 
large trenches, some of which now contain glaciers. Its 
upper parts have a comparatively mature topography, 
and seem to constitute remnants of an ancient pene- 
plain, which has been bodily uplifted, with some disturb- 
ance of original horizontality. The plateau resulting 
from this uplift was deeply trenched along the main lines 
of drainage, and the valleys thus opened were modified in 
characteristic manner by Pleistocene glaciers. They are 
now U-troughs, and some are partly submerged, so as to 
constitute fiords. On the side facing Kachemak Bay and 
Cook Inlet, it was evident that the old glaciers extended 
beyond the position of the modern coast line, but nothing 
was seen to indicate their outer limits. 


Kadiak Island is loo miles long and 50 miles broad. Its 
longer axis, trending northeast, is parallel to the neigh- 
boring coast of the Alaska Peninsula, from which the is- 
land is separated by Shelikof Strait, 30 miles wide. Afog- 
nak Island, a close companion of Kadiak, continues its 
northeasterly trend; and the Barren Islands serve as 
physiographic stepping stones to connect the group with 
the axis of uplift following the oceanic side of Kenai 

The island is mountainous throughout, but contains no 
lofty range. As on the Kenai Peninsula, the summits 
tend toward uniformity and an even sky-line, and there 
are remnants of an ancient uplifted peneplain. Such 
lowlands as we saw are of moderate extent and uneven 
surface. The coast line is sinuous, and some of the nar- 


row bays may properly be designated fiords, although 
less strongly characterized than those of southern Alaska. 

Opportunities for personal observation included near 
views of the northern coast, a brief landing near the west- 
ern extremity, and a longer stay near the eastern extremity, 
with Kadiak Harbor as a base of operations. 

Our landing at the west was on a part of the coast fa- 
cing Shelikof Strait, west of Sturgeon River. The view 
was limited by a fog, but I was able to recognize two 
narrow U-troughs with simple contours. The ridge be- 
tween them, composed of granite rock, was seen to have 
a narrow, straight crest; and a parallel ridge, less clearly 
revealed, appeared to be of the same character. The 
troughs were evidently shaped by glacial ice, and the nar- 
rowness of the intervening crest indicates that the chief 
work was done by valley glaciers, rather than by an over- 
riding ice-sheet. 

Thirty miles farther east we entered the mouth of Uyak 
Bay, a long inlet heading south of the middle of the island 
but opening northward. The mouth lies among hills or 
low mountains, whose thorough rounding indicates com- 
plete flooding by Pleistocene ice. Toward the interior I 
could see mountains, several thousand feet high, whose 
blunt summits told of ice-scoring, and beyond them loftier 
peaks with angular crests. The slopes bordering the bay 
descend steeply to the water, and there is no foreland, but 
the sweeping curves characteristic of the typical fiord are 
wanting. No accumulations of drift were seen. 

Steaming north from Uyak Bay and then eastward, we 
passed two large projections of the coast, one a peninsula 
(Ugat) and the other an island (Uganuk), and the extrem- 
ities of these were thought not to be glaciated. On the 
peninsula are hills with ragged summits, apparently crested 
by outcropping igneous dikes. 

Straits separate the north end of the island from three 



smaller islands, all mountainous. Traversing the straits, 
we found familiar signs of ice work : on Raspberry Island 
a straight wall with hanging valleys; on the Kadiak side 
a general rounding of all summits up to the clouds, which 
hung at about 2,000 or 2,500 feet. Here, too, the water 
is bordered in places by a lowland or terrace (fig. 85), 
carved, for the most part, from vertical slates. In detail 
the lowland is uneven, and it is locally broken into islands, 
but its general plane is easily traced and, as already noted 
by Dall, ^ inclines from east to west. The height ranges 
from about 100 feet to sea-level. 

At the extreme east lies Chiniak Bay, a broad opening, 
invaded on one side by mountain promontories, and partly 


sheltered from the ocean by a group of low islands. Close 
to the islands is the village of Kadiak. The general trend 
of promontories and islands is northeast-southwest. The 
islands are shredded remnants of a plain carved from ver- 
tical slates, probably a base-level plain contemporaneous 
with the terraces along the straits (fig. 86). Their uneven 
surfaces include rounded hills about 100 feet high, but the 
plane of the original peneplain must pass above these. 

All surfaces about the bay are glaciated. The island 
topography is moutonnee, with a large pattern, individual 
bosses being sometimes half a mile or more in length. A 
few patches of glacial polish and striae were found, though 
such records have been generally obliterated by weather- 

» Seventeenth Ann. Rept. U. S. Geol. Survey, part i, p. 863, 1896. 



s I 

fa tJ 

tf} a 

g I 

at ^ 


2 « 


ing. The valleys be- 
tween promontories 
of the mainland are 
U-troughs. The steep 
hill back of the village, 
a mass of slate, is 
smoothed and fluted 
on a grand scale, its 
original topography 
being so completely 
remodeled that its 
drainage seeks new- 
routes and is engrav- 
ing narrow canyons 
across the rounded 
slopes (fig. 87) . 
Northwest of this hill 
stands a higher ridge, 
where the upper limit 
of ice-rounding is seen 
to be about 3,000 feet 
above tide, and where 
several hanging val- 
leys overlook a finely 
sculptured trough (fig. 
88). It is evident that 
a confluent ice-sheet, 
enveloping all but the 
highest summits, here 
flowed to the north- 
east, with a thickness, 
along the present coast 
line, of 2,000 to 3,000 
feet. No important 
masses of drift were 



While these observations cover but a small part of the 
island, they are so distributed as to throw considerable 
light on Pleis- 
tocene con- 
ditions. The 
glaciation of 
the eastern, 
northern and 
western ex- 
tremities of 
the island, and 
the notable 
height to 
whi ch ice 


Showing glacial sculpture and subsequent erosion by a stream. 

sculpture extends at the east and north, indicate a practi- 
cally continuous glacial envelope, from which scattered 

peaks pro- 
jected as 
nuna t aks. 
Afognak and 
the other im- 
portant is- 
lands of the 
group were 
probably in- 
cluded in the 
same envel- 
, ope, and the 
' ice extended 
in many di- 
rections be- 
yond the po- 
sition of the 

present coast line. The apparent absence of glaciation 
from salient features of the northwestern coast indicates 



The lower ground was shaped by a large glacier moving from left 
to right. 


that the ice-sheet of the island was not an overflow from 
the mainland, for mainland ice could not have crossed 
the island without burying deeply the whole northwest 
coast. And the same interpretation may be given to the 
narrow-crested ridge observed between glacial troughs at 
the west, for a strong overriding current would have flat- 
tened the crest. The relation of land to sea in Pleistocene 
time is not shown, but remnants of a low base-level plain 
indicate a pre-Pleistocene period of stability, during which 
the attitude of th6 land was slightly different from the 
present, the eastern side of the island being somewhat 
lower than now. 

The reasoning tending to show that Pleistocene glacia- 
tion was associated with a low sea-level applies to the south 
coast of Kadiak with even more force than to the district 
of Prince William Sound, for the island, standing forward 
beyond the general line of the Alaska coast, is specially 
exposed to the influence of the great ocean current. 


Fragmentary as were our observations in the districts 
we have already described, they were still more fragmen- 
tary in those touched farther west and north, and the pres- 
ent is therefore a convenient point for retrospect and 
summary. The district of the Alexander Archipelago, 
the district of high mountains, the district of Prince Wil- 
liam Sound, Kenai Peninsula, and Kadiak Island, circle 
about the Gulf of Alaska. A curved line passing through 
them is more than i,ooo miles in length, and its extremities 
are 900 miles apart. 

In the high mountain region the Pleistocene glacial 
system seems to have included alpine glaciers similar to 
those of the present time but larger; and these united in a 
system of piedmont glaciers, or possibly in a confluent 
piedmont glacier, which everywhere reached the sea. In 


the other districts the Pleistocene glaciers were confluent, 
extended beyond the line of the present coast, and prob- 
ably reached the sea, although their limit is undetermined. 
In the high mountain region there has been post-glacial 
uplift of the mountains, and in connection with that uplift 
great moraines, which were probably formed at the water's 
edge, have become part of the land. In the other districts 
the apparent absence of similar great moraines is pro- 
visionally explained by the hypothesis that the sea surface 
then lay lower with reference to the land and that the 
subsequent submergence of a portion of the land has con- 
cealed the zone of morainic deposit. 

Assuming that a change has transpired in the relation 
of land and sea since the epoch of maximum glaciation, it 
is of interest to inquire whether that change was a sinking 
of the land or a rising of the sea. The general theory of 
the subject has not yet reached such a condition as to 
aflbrd a satisfactory answer, but, on the contrary, is so 
unsettled as to find advantage in every local determina- 
tion of the nature of the change which may be made on 
independent grounds. In the case of the high mountain 
district, the geologic recency of orogenic movement is 
indicated in other ways, and the post-Pleistocene emerg- 
ence of land is therefore referred with confidence to land 
change rather than water change. In the other districts 
we have no evidence of recent orogenic change, and the 
mountains appear to have been produced by the dissec- 
tion of broad plateaus. The uplands about the Alexander 
Archipelago and Prince William Sound, and those of 
Kenai Peninsula and Kadiak Island, all contain traces of 
uplifted peneplains, the uplift having occurred so long ago 
that the resulting plateaus were profoundly sculptured 
before the advent of Pleistocene glaciers. The districts 
of the Alexander Archipelago and Kadiak Island are also 
characterized by peneplains near present sea-level; and 


low terraces observed in Prince William Sound are pos- 
sibly of the same order. It is, furthermore, probable that 
the pre-Pleistocene dissection of each area was continued 
in association with a comparatively low base-level. In 
view of the large number of common elements, the whole 
region, with the exception of the high mountain district, 
may be provisionally regarded as a unit in its later geo- 
logic history. The uplifted peneplains do not all stand at 
the same height, and there are important differences of 
altitude within individual plateaus. These local differ- 
ences suffice to show that not all changes can be ascribed 
to the sea, and make it probable that the plateaus were 
created by changes originating within the earth's crust. 
The low peneplains also show minor discordances, and 
while these also must be ascribed to crustal movement, 
the fact that they are of moderate amount is, on the 
whole, indicative of general crustal stability. It is a note- 
worthy, and probably significant, fact that the oceanic 
base-level of the region, after resting for a long time at 
the height indicated by these low peneplains, dropped 
below them at the time of the excavation of the fiords 
and then returned to approximately the same position. If 
this oscillation was an oscillation of the land, it was of the 
broad or epeirogenic type, and the association of wide 
extent with approximate uniformity of position after the 
completion of the cycle is most remarkable. It would 
seem probable that so great a general movement of the 
earth's crust would afford opportunity for the relief of 
local strains, and thus be accompanied by important 
differential movements. On the other hand, sensible uni- 
formity for any region of the magnitude here considered 
would theoretically be a characteristic of an oscillation 
of the sea. The local evidence, therefore, seems to me 
more favorable to the hypothesis that the sea was low 
during the fluvial and glacial erosion of the fiords and has 


since risen, than to the hypothesis that the land was then 
high and has since subsided. 


On two occasions we spent a few hours at Dutch Har- 
bor, making short excursions in the immediate neighbor- 
hood, and we also sailed along the north coast on a foggy 
day. The opportunities for observation thus afforded 
were much more limited than those enjoyed by Russell,^ 
and my notes are chiefly of service as affording verifica- 
tion of his description. The north coast, west of Cape 
Cheerful, is faced by a high sea cliff* which testifies to 
rapid aggression by waves. The cliff" shows in cross- 
section a number of U-shaped valleys, and these, so far as 
the fog permitted us to see them, have the simple contours 
characteristic of complete adjustment to the conditions of 
ice flow. Several of them end hundreds of feet above the 
sea, this condition being manifestly due to truncation by 
the receding shore cliff". One of them reaches the shore 
at tide-level, and the walls of that one seemed to be 
sheathed with a layer of drift in which post-glacial rills 
and brooklets have cut narrow gashes. The shore cliff* 
also truncates, at rather high levels, a few V-shaped 
gorges, and the association of these with the glacial troughs 
gave the impression that the Pleistocene glaciers were of 
alpine type and not confluent. As the mountains were 
concealed by fog, I was unable to observe the cirques of 
which Russell makes mention. 

The forms of the hills about Unalaska Bay are not typi- 
cally glacial, but, on the other hand, they are not con- 
structional (with a single exception), and if products of 
atmospheric waste, they are of unusual type. The single 
constructional form is a young volcanic cone near Cape 
Cheerful. The other hills are also of volcanic rock, but 

*Bull. Geol. Soc. Amer., vol. i, pp. 138-140, 1890. 


give little suggestion of the original mountain forms from 
which they were derived. They are irregular alike in 
their larger and smaller features. 

If the forms of land in this part of Unalaska Island were 
constructional, the sinuosity of the coast might be ascribed 
to irregularities of volcanic eruption ; but as they are ero- 
sional, the deep embayments between steep-sided points 
and islands, and the dearth of plains near sea-level, point 
toward a somewhat recent subsidence of the land or rising 
of the sea. 

Spurr describes a series of terraces near Unalaska Bay, 
ranging up to a height of 1,500 feet, ascribes them to 
marine action, and infers a gradual rising of the land in late 
Pleistocene time.^ He notes also that the village of Iliu- 
liuk stands on a spit which is evidently of recent formation 
but is shown by its vegetation to be above the reach of 
storm waves, and infers that elevation of the land is now 
in progress. The last mentioned observation I was able 
to verify, but I was not satisfied that any higher terrace I 
saw had been formed by the sea. 


The most extreme and contrasted opinions have been 
advanced with reference to the Pleistocene condition of 
Bering Sea. It has been stated by one high authority - 
that the western coast of Alaska, the eastern coast of 
Siberia, and various islands of Bering Sea, are all glaciated 
in such a way as to indicate the occupation of the eastern 
part of the sea by an ice-sheet; and it has been asserted 

I Eighteenth Ann. Rept. U. S. Geol. Survey, part in, pp. 266-267, 273, 276, 

•John Muir. On the Glaciation of the Arctic and Subarctic Regions visited 
by the U. S. S. Corwin in the year 1881. Rept. Cruise of the Corwin, 1881, 
Washington, 1885. 


by another high authority^ that there are no evidences of 
glaciation, either general or local, on these various coasts 
and islands. A third investigator,^ also of high rank, 
ascribes the fiords of the Siberian coast to glaciers, but 
finds no evidence of glaciation on the neighboring coast 
of Alaska about Port Clarence. My own opportunities 
for observation were limited to a few hours each on St. 
Paul, St. Matthew and Hall islands, a few hours sailing 
past the Siberian coast, with a brief landing in Plover Bay, 
a distant view of Cape York, a point southeast of Cape 
Prince of Wales, and a few hours on the tundra near Port 
Clarence. The scanty facts thus gathered can not be 
expected to settle the vexed question; but, in view of the 
wide diversity of existing opinion, it appears worth while 
to make record of even hasty observations and first im- 

Of St. Paul Island we saw the southern peninsula. The 
land is there composed of remnants of volcanic cones 

Shows projections which would not survive glaciation. Photograph by U. S. Coast Survey. 

whose softened profiles indicate long-continued weather- 
ing. The forms are smooth, except where cut by the sea 

» W. H. Dall. Bull. 84 U. S. Geol. Survey, p. 258, 1892. Alaska and its Re- 
sources, pp. 461-464, 1870. 

2 A. E. Nordenskiold. The Voyage of the Vega. New York ed., pp. 569, 

583-585» 1882. 


or varied by traveling dunes of lapilli. To my eye they 
conveyed no impression of ice sculpture; I saw neither 
the long parallel grooves and ridges which glaciers some- 
times carve from homogeneous rocks, nor the groups of 
moutonnee bosses which they usually develop where rocks 
are of varied texture. On the main part of the island are 
younger cones with well-preserved craters, and photo- 
graphs show that with these are associated crags such as 
an overriding glacier would not spare (fig. 89). Neither 
is it to be supposed that the craters themselves would 
survive the erosive action of a great ice-sheet. It may 
be affirmed with confidence that if the island was ever 
traversed by a glacier the crater-bearing cones are of 
later origin. 

St. Matthew and Hall islands are also volcanic, but 
without constructional forms. The period of eruption was 
so remote as to give time for the complete subsequent re- 
modeling of the surface by weathering and erosion. The 
coast shows a succession of cliffs, with rare bays and 
spits, and is evidently retreating rapidly before the attack 
of the waves. The higher slopes, though sometimes 
steep, are in general mature, and well adjusted to the con- 
ditions of erosion in a climate which obstructs the flow 
of water by clothing all surfaces with a sponge-like mantle 
of mossy and herbaceous vegetation. I saw nothing of 
the peculiar forms characteristic of glacial sculpture, but 
noted, on the contrary, a few blunt pinnacles projecting 
from the general surface and exhibiting such ragged de- 
tails as one does not find in glaciated regions. Figure 90 
represents one of these which happened to come within 
the field of a photograph. 

The ordinary landing at Port Clarence is upon a long 
spit, but we visited the mainland also, going ashore at a 
point where a gently undulating surface rises within a 
few miles to hills several hundred feet high. The rock is 



a highly inclined slate, and the shaping of the surface has 
been wholly by erosion. Except at the coast, the rocks 
are concealed by tundra. This spongy growth obstructs 
the flow of water, so 
that streams are rare ; 
but we landed at one 
of these rare streams 
and had a view of its 
valley (fig. 91). The 
valley is evidently 
one of mature de- 
velopment, and its 
profiles are perfectly 
adjusted to the associated lines of drainage. The divides 
are broadly rounded, but the rounding is that characteristic 
of inter-stream summits where the vegetal mat is close, 
and is distinctively non-glacial. The bed-rock here has 
the same physical character as that in the Kadiak region, 
but the topographic aspect is altogether different. About 

FIG. 90. 


■-v<*'-*^ .X "■;>-' ^"^Kii>ji;;*?*cs:i:--*i;:^-£ 


Kadiak the hills have moutonnee forms, the hills and hol- 
lows have a dominant trend, and the drainage is youthful. 
At the Port Clarence locality the topography does not 



exhibit a dominant trend and all slopes are fully adjusted 
in harmony with the drainage. 

The southeastern part of the Siberian peninsula is char- 
acterized by low mountains with spurs projecting seaward 
as promontories and alternating with fiord-like bays. The 
topographic details near the coast fall into three categories : 
(i) A system of relatively gentle slopes chiefly occupying 
uplands; (2) a system of relatively steep slopes chiefly 
exhibited in the walls of the fiords (at points of junction 
these are sharply contrasted with the slopes of the first 
system) ; (3) coastal features, especially shore cliffs and 

The sky-line follows the boundary between the steep fiord wall and the smooth topogfraphy 
of the upland. A spit projecting from the end of the wave-wrought shore cliflf protects the 
more distant part of the wall. 

spits, the product of wave action under present conditions. 
The smoother and older topography is not altogether de- 
void of steep slopes, but gives an impression of close ad- 
justment between processes of subaerial erosion and the 
unequal resistances of rock masses. I saw nothing in its 
profiles and contours to indicate glaciation. The newer 
and steeper slopes are associated with the troughs con- 
taining the bays in such way as to suggest glacial action, 
but those that we passed near betrayed no smoothing, 
grooving, or other minor feature of glacial abrasion. The 
shore walls of Plover Bay are precipitous rock cliffs at top 
and consist of talus at base, one phase passing into the 
other in a manner suggesting that the original rock pro- 
file was somewhat similar to the one brought about by 



partial disintegration. They run straight for long dis- 
tances. The chief agencies competent to produce such 
features are faulting and glacial sculpture, and in this 
case glacial sculpture appears to me the more probable 
agent, although subsequent weathering seems to have de- 
stroyed those minor details of configuration which one 
naturally seeks as confirmatory evidence. 

There are two accessory features which lend support to 
the hypothesis 
of glaciation. 
In the far dis- 
tance, at the 
head of the 
bay, we could 
see that its 
trough is con- 
nected with 
two or more 
land valleys, 
and it was evi- 
dent that the 

valley most nearly in the direct line of prolongation is dis- 
tinctly U-form in cross-profile and has walls of simple con- 
tour. The other feature is a niche, high on the wall of the 
fiord, having the form of a cirque or hanging valley (fig. 93). 
In this case I attach little weight to the testimony of the 
hanging valley, because it has no companion on the long 
line of cliffs, and therefore may possibly be a simulative 
form, determined by local peculiarities of rock texture; 
but the distant valley is distinctively glacial in habit. 

It seems to me on the whole probable that the fiords 
of this coast contained Pleistocene glaciers of large size, 
which extended farther seaward than the general line of 
the present coast, but that the spaces between the fiords 
were not covered by ice. 

Figure 92 gives an oblique view of the same wall. 


My observations on the coasts of Bering Sea may be 
summed in the statement that Plover Bay and neighbor- 
ing Siberian fiords have features indicating local glaciers 
of considerable magnitude, that evidence of glaciation was 
seen at no other points, and that certain crags and pin- 
nacles on St. Matthew and St. Paul islands seemed incon- 
sistent with the theory of a continental glacier in the 
Bering Sea region. My interpretations at the north agree 
substantially with those of Nordenskiold; at the south 
with those of Russell. 

So far as the Port Clarence region is concerned, what 
I have said above has become ancient history before reach- 
ing the press. A delay of four years between observation 
and publication is fatal to novelty, if one's theme concerns 
a region developing under the stimulus of the discovery 
of gold. Near where we landed on the shore of Port 
Clarence the town of Bering now stands, and all Seward 
Peninsula has been explored by the prospector. To aid 
him the U. S. Geological Survey has sent active parties 
of geologists and topographers; and as the proof sheets 
of these pages pass through my hands, I am able to 
examine contour maps of a large part of the peninsula, 
and study three comprehensive reports of geologic recon- 
naissance. These reports are by Brooks, Mendenhall and 
Collier, and tell of explorations and surveys made in the 
seasons of 1900 and 1901.^ 

They cover the general question of Pleistocene glacia- 
tion in a demonstrative and altogether satisfactory way. 
The Kigluaik Mountains, between Port Clarence and 
Cape Nome — mountains with an extreme height of about 

1 A Reconnaissance of the Cape Nome and adjacent Gold Fields of Seward 
Peninsula, Alaska, in 1900. By Alfred Hulse Brooks. 1901. See pp. 42-53. 

A Reconnaissance in the Norton Bay Region, Alaska, in 1900. By Walter 
Curran Mendenhall. 1901. See p. 208. 

A Reconnaissance of the Northwestern Portion of the Seward Peninsula, 
Alaska. By Arthur J. Collier. In press. See pp. 24-29 and 34-42. 


4,700 feet — nourished in Pleistocene time local glaciers 
of some magnitude. These are attested by U-troughs, 
cirques, moraines, and moraine lakes. One of them 
passed southwestward beyond the foothills of the range, 
and may have reached the sea. Another approached or 
reached sea-level at the north. The Bendeleben Moun- 
tains, farther inland, also contained glaciers, but too small 
to push beyond the foothills. Among the York Moun- 
tains, which stand between Port Clarence and Cape 
Prince of Wales, and have an extreme height of about 
2,900 feet, were probably small glaciers, but, if so, they 
were wholly contained in the mountain valleys. Except 
for these local developments, the surveyed parts of 
Seward Peninsula — namely, the southern and western 
parts — were not occupied by Pleistocene ice. Over 
large areas the mantle of residuary waste lies undisturbed 
on the rock from which it was derived; and in these areas 
are angular and slender crags, as well as perched boulders 
of disintegration, lying in their original positions. Other 
large areas bear rolled gravels, associated with a series 
of marine terraces. 

One of the marine terraces described by Collier and 
Brooks was seen by us along the southern base of York 
Mountains. It is there a conspicuous bench, ending sea- 
ward in a steep bluff, and has a height of about 600 feet. 
According to Collier, its uplift has been unequal, so that 
the old marine plane is now a warped surface. The fact 
of warping also accords with my observation, for such 
parts of the neighboring Siberian coast as I was able to 
Study from the ship seemed altogether free from marine 

An observation of progressive modern change of level 
was made on St. Matthew Island. A small bay on the 
east shore, near Glory of Russia Cape, has been cut off 
from the sea by a series of shore bars. After the first bar 


or Spit was completed, another was built outside it, and so 
on to the number of six or more. The one next the sea 
has a crest considerably above high-tide, at the extreme 
limit reached by storm waves, and the top of it is covered 
by drift-wood, to which annual additions are doubtless 
made. The next bar, lying back of this and parallel to 
it, is six feet lower. It also is covered by drift-wood, but 
the wood is decayed, so that no sound logs were found. 
Two others are successively lower and have no drift- 
wood, and the innermost are so low as to be covered by 
the water of the bay. It is evident that each of these 
ridges of shingle was formed by storm waves at the shore, 
and we may assume that its crest height was originally as 
far above high-tide as the crest of the outer ridge is now. 
The existing differences in height have resulted from the 
gradual sinking of the land, and a rude indication of the 
rate of sinking is given by the drift-wood. The inner 
ridges were made so long ago that the drift-wood they 
originally bore has rotted away and disappeared. The 
age of the second ridge has given time for only the partial 
decay of the timber upon it, and the inference is that the 
island has sunk to the extent of six feet in a period less than 
that necessary for the complete destruction of logs through 
the processes of decay. 




Some of the field observations recorded in the preced- 
ing chapters have more than local interest in that they 
open theoretic questions and lead to suggestions bearing 
on the general subject of glaciers and their work. Certain 
of the suggestions have already been noted, as they seemed 
to belong to the discussion of the associated observations, 
but others may more appropriately be considered by them- 
selves and have been reserved for the present chapter. 
The general considerations already set forth refer to an 
annual cycle in the distal extent of tidal glaciers (p. 22), 
the production of features of shore topography by waves 
generated by the calving of icebergs (p. 69), anomalies in 
the variations of glaciers (p. 106), the origin of pitted 
plains (p. 54), and the origin and interpretation of hanging 
valleys (p. 114). The considerations to be presented in 
the following paragraphs pertain (i) to the broader char- 
acters of the surface of a glacier, (2) to the conditions af- 
fecting types of glacial sculpture, (3) to the conditions 



of wear below sea-level, and (4) to the parallelism of gla- 
ciers and rivers. 


Evenness as Compared to Rock Floor, — The parts of 
glaciers which came under my observation were the lower 
or distal portions, with surfaces usually less than 1,000 
feet above sea-level. In these lower parts the master 
characters of the surface are a forward slope in the direc- 
tion of flow and horizontality in the direction normal to 
the flow, the direction of flow being inferred from the 
courses of medial moraines. 

The configuration of the rock floors beneath the glaciers 
could not be directly observed, but it was possible to infer 
their general characters, with high probability, from what 
could be seen of the rock floors in front of the glaciers. 
Such rock floors are in all cases abandoned glacier beds, 
having been covered by the ice not only in Pleistocene 
times but, in many instances, in historic time also. As 
a rule these bared portions of the glacier troughs exhibit 
much irregularity. The fiords vary rapidly in depth; in 
places they are diversified by islands. The land troughs 
have hills and hollows; and other hills jut through the 
glaciers as nunataks. It is therefore believed that the 
bottoms of the glacier channels have in general consider- 
able inequalities. 

These irregularities are only slightly represented in the 
configuration of the glacier surface. The greater inequali- 
ties of the longitudinal profile of the bed are shown by 
cascades of the glacier, but inequalities of the cross-pro- 
file are rarely indicated by visible shapes of the ice, and 
bosses or hills hundreds (perhaps thousands) of feet high 
may fail to influence the surface. If the summit of a boss 
approaches the surface it produces crevasses, but the 
broader features of the surface contours are not changed. 


This leveling of the surface is of course ultimately 
caused by gravity, and it is immediately brought about by 
variations in the flow of the ice, the directions and veloci- 
ties of the flow of different parts being so modified as to 
permit the ice to pass around and over obstructions with- 
out lifting the glacier bodily. The existence of internal 
variations in direction, being attested by the arrangement 
of striae on lands that have been glaciated, is a well-recog- 
nized fact, but variations of direction alone are not suffi- 
cient to explain the coexistence of an approximately even 
glacier surface with a very uneven glacier bed. The con- 
dition of continuity can not be satisfied without variations 
of velocity also within the mass. The adjustment of an 
ice stream to an irregular channel through the formation 
of differential currents is thus quite analogous to the ad- 
justment observed in water streams, although the greater 
viscosity of the ice may be assumed to prevent the occur- 
rence of reversed currents or eddies. 

Another factor in the leveling of the ice surface is prob- 
ably connected with ablation, or the process of wasting by 
melting and evaporation. Portions of the Muir and Hugh 
Miller glaciers which were demonstrably stagnant, never- 
theless exhibited to the general view a conspicuous even- 
ness, and the same remark applies to glacier remnants 
stranded, like Reid's * Dying Glacier,' on saddles, or trough 
summits. These motionless and wasting ice bodies, 
though not tabular but curved downward toward their 
edges, were bounded by simple symmetric contours, in- 
dicative of equable reduction by the wasting agents and a 
general interdependence of process for the whole surface. 

Lateral Cliff. — Coexistent with the general tendency 
toward evenness are several kinds of unevenness, each 
determined by some evident special condition. Wherever 
the side of an ice stream was observed adjacent to a rock 
wall, it was found to present a cliff* toward the rock, the 


ice cliff and rock wall constituting the sides of a narrow 
valley or fosse, usually 50 to 100 feet deep. A stream of 
water sometimes followed the valley. This feature, fa- 
miliar in all glacier districts, has been explained as due to 
the heat acquired by the rock through insolation and then 
conveyed by radiation to the adjacent ice; and the stream 
of water, when present, would help to account for the 
valley, for so much of its volume as came from the sun- 
heated rock would be warmer than the water of ablation 
and have some power to melt ice. 

Crevasse Cycle, — Wherever the work of the sun is not 
complicated by the presence of rock debris, the inequali- 
ties initiated by crevassing are carried by ablation through 
a regular cycle of change, ending in their complete re- 


The lower part of the ice is undivided ; the upper is split into slabs and columns. The 
dark hill in foreground is of ice with a cover of gravel, a remnant of the retreating glacier. 

moval. In the crevassing which begins abruptly at the 
head of a cascade, the cracks divide the ice into flat- 
topped, elongated blocks, usually tapering toward the ends 
and more or less connected at the surface by slender 
masses analogous to the slivers of half-broken timber. 
Whatever the distance downward to which the cracks may 
originally extend, the resulting permanent crevasses have 



only moderate depth, the ice being welded into a practi- 
cally continuous mass beneath. Wherever a crevassed 
tract was exhibited in section in a tidal cliff, the crevasses 
were seen to terminate uniformly along a definite zone, 
which was usually nearer the top of the cliff than its base. 
The depth of this zone was estimated in different instances 
at from 50 to 125 feet, but this did not represent the full 
original depth of the crevasses, as something had in every 
case been lost by ablation (fig. 94). 

As soon as the cracks are opened, melting begins on 


their faces, the rate being greatest above, and the flat 
tops of the ice blocks, the seracs of alpinists (fig. 95), are 
converted into roof-like crests and pinnacles. In this 
condition the surface is nearly or quite impassable (fig. 
96). With continuance of ablation the height of the 
seracs is reduced, their slopes become less steep, and 
many connecting cross ridges become available to the 
wayfarer, so that with the exercise of care and patience 
one can make his way safely in any direction. Figure 
97 shows a characteristic field of this sort, crossed by 
some of our party on their way to the great nunatak of the 
Columbia Glacier. A continuance of the same reduction 
by ablation eventually obliterates the ice waves alto- 



gether, leaving a plain surface over which progress is 
absolutely unimpeded (fig. 98). 


Thus the cycle of ice degradation by ablation is strictly 

analogous to 
the cycle of 
land degrada- 
tion by erosion; 
an original 
plain is rapidly 
converted into 
a region of 
highly acci- 
dented topog- 
raphy, which 
slowly returns 
to the plain condition through the removal of its promi- 
nences. If the cycle is interrupted by the formation 
of a new system of crevasses, the analogy still holds, 

Illustrating gradual smoothing of surface. 




FIG. 98. 


for the features of the new cycle thus instituted at first 
combine with those of the old and eventually supplant 

As my excursions on glaciers were all short, it hap- 
pened that I never saw a 
complete illustration of the 
ablation cycle on one gla- 
cier, but such examples 
should be readily discover- 
able. Where the even bed 
of a glacier increases its 
grade, where it is interrupted 

by a step, producing a CaS- The rock retards the melting of the ice on 

cade, or where the OVerrid- -^-^ ^^ -«*«. -^ thus preserves a pedestal. 

ing of a submerged peak produces breaking strains in the 
upper part of the ice stream, the smooth ice plain above, 

the prismoidal 
blocks, the acute 
peaks, the gradually 
subsiding waves, and 
the final ice plain 
should appear in reg- 
ular sequence, sub- 

lUustrating the formation and obliteration of crevasses stantiallv aS rCDrC- 
and seracs. *' r 

sented in figure 99. 
An exceptional condition was observed on the tongue of 
the Columbia Glacier which flows into the western embay- 
ment of its valley. A plain surface was there interrupted 
by an extensive plexus of crevasses, which were filled to 
the brim with water and snow (fig. 100). The normal 
cycle appeared to be varied in this case by a lack of 
drainage, surface ablation progressing only where the air 
had access, and truncating the seracs down to the water- 
line. It is conceivable that under such conditions a set 
of crevasses originating from horizontal stresses may pro- 

FIG. 99. 





duce no crests and pinnacles, but remain as mere inter- 
ruptions of a general plain surface until obliterated by 
progressive ablation. 

This exceptional condition draws attention to the fact 
that crevasse systems are normally underdrained. Look- 
ing into crevasses on a warm day, 

one may sometimes see the water 
of ablation in slender rills disap- 
pearing down tunnels or shafts in 
FIG. loo. IDEAL SECTION OF thc compact bluc Icc below — to 

be gathered doubtless in englacial 
or subglacial streams, and eventu- 
ally escape at the end of the glacier. 

Where the ice carries a heavy back-load of drift the 
normal crevasse cycle is greatly modified (fig. loi) and 
its completion indefinitely postponed. 
As already mentioned in connection 
with the Hidden Glacier, the drift 
falls into the crevasses and gathers at 
their bottoms. As the intervening 
ice blocks acquire acute crests the 
greater part of the drift rolls and 
slides from them, and they retain only 
enough to darken the surface. Under 
the familiar law that a sprinkling or 
thin cover of drift promotes melting, 
while a heavy mantle retards it, the 
pinnacles are rapidly reduced, and 
their sites are eventually depressed 
below the driit masses accumulated 
in the crevasses. The original asperi- 
ties made by the crevasses have now 
been destroyed, but a secondary system has been evolved, 
which tends in similar manner to produce a tertiary sys- 
tem, and so on indefinitely. Wherever we found a broad 

Crevasses formed in drift- 
covered ice (a) accumulate the 
drift (b), and eventually be- 
come the sites of hills (c). 


tract of ice heavily loaded with drift, its surface was made 
up of hummocks and hollows, varied here and there by 
cliffs of black ice, down which pebbles and boulders oc- 
casionally rolled in shifting their position from hill to 
hollow. These irregularities, involving, as they do, in- 
equality in the distribution of the drift on the surface of 
the ice, help to account for the irregularity observed in 
terminal moraines. 


The work of rock sculpture accomplished by the middle 
and lower parts of a glacier is performed chiefly by the pro- 
cesses of abrasion and plucking. In abrasion, fragments of 
rock held in the under part of the ice, being dragged over 
the fixed rock of the glacier bed, file away and reduce it. 
In plucking, blocks of bed-rock, being partly surrounded 
by the ice, are forced from their bearings and rolled or 
slidden forward. If the plucked blocks have originally 
stood as projections, they may be broken away, even if 
quite firm and flawless ; otherwise it is probable that they 
can be removed only if originally separated by joints or 
other structural partings. The waste resulting from abra- 
sion is clay and sand; plucking yields boulders. 

One of the chief factors on which the rate of abrasion 
depends is the velocity of the moving ice. If the bed- 
rock surface is uneven, the ice does not flow over all parts 
of it at the same rate, but moves slower in the hollows 
and faster across the prominences. This difference results 
partly from the condition of continuity, which demands 
higher average speed where the cross-section is less, and 
partly from the tendency of parts embayed in hollows to 
lag behind the general mass. The prominences are there- 
fore abraded more rapidly than the adjacent hollows, and 
the profile is thus reduced to simple forms. 

Another factor on which rate of abrasion depends is 


pressure ; the abrasion is more rapid as the pressure of the 
glacier against the bed-rock is greater.^ The resistance 
which the moving ice, through its viscosity,^ opposes to 
change of form, causes it to press unequally on different 
parts of an uneven bed, and to abrade most rapidly those 
parts whose prominence compels the ice to change its 
direction. Thus in a second way there is a tendency to 
reduce the profile of the bed to simple forms. 

The amount of resistance developed by viscosity de- 
pends on the rate of deformation; more force is necessary 
to deflect the ice quickly than to deflect it slowly. The 
parts of the bed which cause the most abrupt turns are 
therefore subjected to greatest pressure and to greatest 
wear, with the result that the profiles of the bed eventually 
become curves of large radius, adjusted to slow bending 
of the moving ice. 

If the general motion of the ice is very slow, the resist- 
ance developed by viscosity is small and the resulting 
sculpture curves have comparatively small radius. If the 
ice moves rapidly, the sculpture curves have large radius, 

1 There are two theoretic limits to the law that abrasion increases with pres- 
sure. It has been argued by N. S. Shaler that because the melting temperature of 
ice is lowered by pressure, the basal part of a thick glacier must consist of water 
instead of ice (Outlines of the Earth's History, pages 237-239, 1898); and such 
'pressure-molten' water would manifestly be powerless to grind rock waste 
against the rock bed. G. F. Becker has suggested to me in conversation that 
where the pressure is great there also the elastic limit of the ice is far exceeded 
and the ice should be expected to flow about rock fragments so as to incorporate 
them in the glacier and reduce or destroy their effectiveness as tools of abrasion. 
These considerations are not included in the above analysis because, while I do 
not see my way to their satisfactory discussion, I fail to perceive that they help 
to explain the phenomena of ice erosion as seen in Alaska. Whatever their in- 
fluence may be, it has not prevented exceptionally great erosion in places where 
the Pleistocene ice was exceptionally deep. 

2 In untechnical usage the word viscosity is ambiguous, being applied to a 
property of liquids opposed to mobility and to a property of solids opposed to 
rigidity. In the present paper it has the technical meaning given by the physi- 
cist, and is the property of fluids and solids in virtue of which internal dif- 
ferential movement, or shear, consumes time. The greater the viscosity the 
slower the yielding to a given shearing force. 


The tendency toward curves of large radius is more effect- 
ive where the material of the bed is easily worn than 
where it is obdurate. 

A third factor affecting rate of abrasion is the material 
abraded. Some materials yield more easily than others 
and are worn more rapidly. Where the material of the 
glacier bed is heterogeneous, including both yielding and 
obdurate rocks, there is a tendency to hollow out the 
yielding rocks and leave the obdurate masses prominent. 
This tendency is opposed by those arising from the vis- 
cosity of the ice, and the type of the resulting sculpture 
in each individual case is a compromise. It is thought 
that the relative importance of viscosity is greater with 
swift-moving ice than with slow-moving ice. 

A fourth factor is found in the quality and quantity of 
the abrasive material, the rock particles set in the base 
of the ice. The particles picked up from shale would not 
be effective in grinding quartzite, but particles from 
quartzite would act vigorously on most other rocks. The 
abrasive action of pure ice is probably nil. The influence 
of this factor is not easily formulated, but there can be no 
question that it qualifies the influence of other factors in 
important ways. 

The sculpture wrought by plucking differs notably 
from that due to abrasion. The plucking of a block of 
rock removes a projection and leaves a hollow. A sur- 
face which has been reduced chiefly by plucking abounds 
in salient and reentrant angles, and would be called 
hackly if its pattern were smaller. Usually its salients, 
and often its reentrants, are rounded by subsequent abra- 
sion, producing a topography to which Saussure's title of 
moutonnee is peculiarly applicable. 

The conditions which locally determine plucking rather 
than abrasion are not clear to me. Evidence of plucking 
is seen more frequently on hard rocks than on soft. The 


clearest examples are on salient masses, but this may be 
merely a question of exposure, for the finest illustrations 
of abrasive sculpture are also on salients. Where a heter- 
ogeneous rock bed acquires an uneven surface by abrasion, 
the prominences of obdurate rock would be specially ex- 
posed to plucking, and it is easy to understand that pluck- 
ing may be combined with abrasion in the reduction of 
such a tract. 

The unevenness produced by plucking is a minor feature 
of the sculpture topography. When greater features are 
considered it is evident that plucking as well as abrasion 
is more active on salient than on reentrant profiles, for how- 
ever hackly an ice-worn hill may be in detail, its general 
profile and contours have the same sweeping curves which 
characterize the products of abrasion. 

The preceding discussion, which for brevity has been 
given somewhat deductive form, is largely based on field 
observation, being the result of an endeavor to understand 
the varied phenomena of sculpture observed in Alaska. 
In a region where the evidence of great glacial erosion is 
overwhelming, where multitudinous hanging valleys, the 
general obliteration of spurs from the sides of U-valleys, 
and the dominant and thorough rounding of crests and 
corners of hills and small mountains, testify to an enor- 
mous amount of glacial degradation, it was a matter of 
surprise to find the reduction of the surface to smooth- 
sweeping curves a somewhat rare phenomenon. By far 
the greater number of well-exposed glaciated areas, even 
where the degradation has been profound, abound in low 
embossments and in more or less angular groins or re- 
entrant spaces showing little trace of abrasive action. 
These surface characters presented themselves as facts 
requiring explanation; and I have come to regard them 
as indications of the great importance of plucking in the 
work of glacial erosion. 

. /') r\\i< 
j'jnonrrncnq rnovz-ODi n/; gv/OfI« bnuoi^^'jioi arlT — .•i^v\"^^N^ '\vir^>A 
■■': ! . ( jioijIO nrQ mb "to dmo« I8u{ ,Yf)llKV liuM to Ihiv/ iau-j t>fb So 

bnfibioV i !i bnfi Je»Inl liuM awod^ ucjntjl^ib -^riT 

111 .1) jniKiorn }nf>bnr, nu — ;ildv/n)> bb;I-i3JKv/ \o 
'V.)8' f(? 'J""(. fJT-»dIK) .vi .O vd j>'.^dqf;T;^»otod*I 

nces oi 


Glaciated Rocks 

Upper Figure . — A portion of the east wall of the valley containing 
Muir Glacier, at a point nearly opposite the end of the glacier in 1899. 
The conspicuous polished surfaces are 200 to 300 feet above tide-level, 
and were probably covered by ice a century ago. The direction of 
movement was from left to right. See page 207. 

The axis of the camera was elevated. Photographed by G. K. Gil- 
bert, June 9, 1899. Negative no. 243, United States Geological 

Lower Figure. — The foreground shows an ice-worn prominence 
of the east wall of Muir Valley, just south of the Dirt Glacier. It is 
viewed from the up-stream side. Its altitude is about 550 feet. It 
was probably covered by ice a century ago. See page 208. 

The distance shows Muir Inlet and its western wall, with a foreland 
of water-laid gravels — an ancient moraine delta. 

Photographed by G. K. Gilbert, June 9, 1899. Negative no. 269, 
United States Geological Survey. 

. L rare i 

ter nurabe U-exposeci 

■ has been pn 


H. A. E. VOL, III 





The importance of plucking may also be inferred, as in 
fact it has been by Dana, from the abundance of boulders 
in the moraines of an ice-sheet. In the waste deposited 
by an alpine glacier it is not easy to discriminate 
plucked boulders from the boulders which have fallen to 
the ice from adjacent rock slopes and been carried for- 
ward as back-load, but the waste carried to the edge of a 
great ice-sheet like the Laurentide has all been picked 
up as well as transported by the ice. A portion of such 
a body of waste must be referred to the mantle of resid- 
uary and alluvial debris found initially on the land by the 
expanding ice mass, and this portion of course includes a 
contingent of boulders; but there is no reason to regard 
this factor as of great importance. It is probably much 
more than offset by the destruction of boulders in the 
glacial mill, for in the making of the rock flour which 
constitutes the body of glacial till, the abrasion of the 
coarser waste carried by the bottom ice may approach, 
or even exceed, the abrasion of the rock floor. While 
these various factors do not admit of definite valuation, I 
think it fair to say that the ratio of boulders, on the one 
hand, to cla}^ and sand, on the other, in the waste deposits 
as a whole, is something less than the ratio of plucking 
to abrasion in the erosive work of such an ice-sheet as 
the Laurentide. 

The study of this subject made such slow progress in 
the field that opportunities for good photographic illustra- 
tion were not improved. The views reproduced in plate 
XVIII were taken to show abrasive work, and illustrate 
plucking only incidentally. The upper view looks up the 
wall of Muir Inlet. It shows stratified rock, with a boss 
of more massive, possibly plutonic, rock beyond. A 
century ago all the stratified rock was covered by Muir 
Glacier. At the left near the sky-line the strata are seen 
to be obliquely truncated, and the plane or nappe of trun- 


cation can be traced across the view. It really includes 
the foreground, at least down to the mantle of moraine; 
and it is part of the wall of the glacier trough, wrought by 
the ice into curves of large radius. But the detail of this 
erosion plane, shown in the center of the field, is distinctly 
hackly, as though (and probably because) the degrada- 
tion of the stratified rock was accomplished largely by the 
breaking away of blocks. The prominent angles and 
edges of strata are abraded and highly polished on their 
stoss sides, but their lee facets are unworn. 

The lower view shows the stoss side of a granite knob, 
also near Muir Glacier. Here the indications of abrasion 
are most conspicuous, but a groin crossing the foreground, 
and a niche at the right, are probably remnants of scars 
made by plucking. 

The rock areas best showing the character of the sculp- 
ture due to plucking are in the barren regions about the 
glaciers, and these were often included in our views of 
glaciers ; but the photographic methods employed to show 
details of the white ice do not secure the details of dark 
rocks. The peculiar embossment topography is merely 
suggested in some cases by the pattern which results from 
the preservation of snow in concavities of the surface. 
See the mountain spur beyond Reid Glacier in figure 14 
and the hill at the left of Hugh Miller Glacier in plate iii. 
Similar patterns appear on both sides of Serpentine Glacier, 
in the photogravure at page 124 of volume L 

Figure 102 contrasts several types of sculpture. A tract 
bordering the water of the fiord at the right shows elab- 
orate fluting on a grand scale, the abrasive work of a 
powerful, and doubtless fast-moving, ice stream, flowing 
from left to right through the fiord. The material here 
sculptured is argillaceous slate, similar to that at Kadiak 
(figs. 86 and 87). The spurs beyond are partly of more 
obdurate material, including schists and granitoid rocks; 



they were sculptured chiefly by currents approaching the 
fiord from the right; and plucking was an important fac- 
tor of the process. The peaks against the sky were above 
the Pleistocene ice. 

In the last example the unevenness in the region of 
extensive plucking is partly due to the varying resistance 
of heterogeneous bed-rock. This factor finds strong ex- 
pression in the forelands and low islets at New Metlakatla, 
Sitka, and Kadiak, where peneplains well reduced toward 
base-level, and therefore 
presumptively of nearly 
even surface, were much 
roughened by a moderate 
amount of glacial deg- 
radation (see figs. 64, 65 
and 86, and pi. xvii). 

It is also strongly ex- 
pressed on the slopes of 
Unalaska Island about 
Unalaska Bay. There is 
a general lumpiness of 
the surface which would 
throw doubt on the 
theory of extensive Pleis- 
tocene glaciation were 
the theory not strongly 
supported by the con- 
spicuous glacial sculp- 
ture of other parts of the 
island. Except where 
exposed in sea cliffs, the rocks are largely covered by a 
tundra mat, and this leaves room for doubt as to the steeper 
slopes, where many of the knobs maybe landslips; but on 
the gentler slopes the knobs are clearly obdurate elements 
of the heterogeneous volcanic mass, rendered prominent by 



differential erosion. A strip of lowland between Dutch 
Harbor and Iliuliuk is so hummocky and so set with lake- 
lets as to remind one of a terminal moraine of the Lauren- 
tide ice-sheet, and I half expected to find it a heap of glacial 
waste; but every discovered break in its turfy mantle 
revealed volcanic rocks in situ, and I was forced to regard 
its undulations as products of sculpture. 


The discussion, in the last chapter, of the fiords of the 
Alexander Archipelago assumes (page 163) that a tidal 
glacier is partly supported by the sea, so that the full 
weight of the ice does not press on the rock floor, and 
the glacier's power to erode is correspondingly diminished. 
This assumption has been often made, and is usually given 
quantitative form. The sea is said to sustain a portion of 
the glacier equal in weight to the body of water displaced 
ct by the ice, and correspondingly to dimin- 

ish the pressure of the glacier on its bed. 
When the subject is approached in a 
certain way this view seems altogether 
plausible. In figure 103, a represents in 
section a sea 2,000 feet deep with a flat 
bottom. In the sea floats an iceberg of 
which the sides are vertical and the top 
and bottom are horizontal planes. Its 
thickness is 1,600 feet, and the submerged 
part measures 1,400 feet, the densities of 
the ice and the water having the ratio of 

FIG. 103. DIAGRAMS ^ r^t • i 

ILLUSTRATING FLO- ? to 8. Thc watcr sustams the entire 
TATioN THEORY OF wcight of thc Icc. Now couccivc the 

TIDAL GLACIERS. ^ r .^ ^ I. J • J 

water of the sea to be dramed away. 
The block of ice sinks to the bottom {b) and is wholly 
supported by it. Again, conceive the water of the sea to 
be only partly withdrawn, the depth being reduced from 


2,000 feet to 1,400 feet {c). This is equivalent, so far as 
the block of ice is concerned, to raising the sea bottom 
from its position in a until it touches the base of the ice- 
berg. The ice is not lifted by the sea bottom; it still pro- 
jects into the air just 200 feet; it is still supported by the 
water, and though touching the sea bottom does not press 
on it. Finally, conceive the water drawn down until its 
depth is but 700 feet (d). This depth of water is just 
able to float a berg 800 feet thick. Therefore 800 feet of 
ice, or one-half the thickness of the block, are now sup- 
ported by the water, and the remaining 800 feet by the 
sea bottom. 

Let us now approach the subject in a different way. 
Begin with a block of ice of the same dimensions as be- 
fore, resting on a horizontal bed, with which it is every- 
where in contact (fig. 104, a). The pressure on each square 
inch of the bed equals the weight of the 
column of ice above it — about 640 
pounds. Now introduce sea water about 
the ice until it has a depth of, say, 700 
feet (d). The water presses horizontally 
against the vertical faces (as indicated by 
the arrows), but, as there is no vertical fig. 104. diagrams 
component to a horizontal force, the water illustrating non- 

. , ,.^ , . 111 FLOTATION THEORY 

pressure neither lifts the ice block nor ^^ ^^^^^ glaciers. 
pushes it down. The block continues 
to rest on the bed, exerting still a downward pressure of 
640 pounds per square inch of base. This line of reason- 
ing seems quite as plausible as the other, but the result 
is different. 

A little consideration discovers the cause of the dis- 
crepancy. In the first analysis it is tacitly assumed that 
the water exerts its pressure not only on the sides of the 
ice block but on its base, and this whether the block floats 
free or touches the sea bed. In the second analysis it is 


tacitly assumed that the contact of the ice with the bed 
excludes the water. In order to determine which analy- 
sis is applicable to the case of the tidal glacier, it is neces- 
sary to consider whether the nature of the contact between 
the glacier and its bed is or is not such as to exclude the 
sea water and its pressure. 

It was suggested by my colleague G. F. Becker that a 
laboratory test might be applied to one of the principles 
appealed to in the second of the two analyses — the prin- 
ciple that a solid rectangular block immersed in a liquid 
is not buoyed up by the liquid provided its base is in com- 
plete contact with the bed on which the liquid rests; and 
at his request two pertinent experiments were made by 
A. L. Day in the physical laboratory of the U. S. Geolog- 
ical Survey. In the first experiment a small slab of plate 
glass was cemented to the bottom of a glass vessel (fig. 
105), for the purpose of giving an accurately plane surface, 

and the vessel was then partly filled 

with mercury. A second piece of 

plate glass was immersed in the 

mercury and pushed down until 

one of its faces came into contact 

FIG. lo^ CONTACT PHENOM- wlth thc facc of thc fixcd slab. As 

ENA OF GLASS AND MER- thc dcuslty of mcrcury is about five 

. t-1 , * cuRY. ^ times that of srlass, some force was 

A block of glass rests on the bot- ^ ^ 

torn ; a similar block floats nCcdcd tO immCrSC thc block of 

at the surface. i . 1 j 

glass, but as soon as contact had 
been secured with the slab below, the block remained at 
the bottom. Not only did it show no tendency to rise, 
but force was necessary to detach it. As both glass sur- 
faces had been carefully cleaned, there could be no cemen- 
tation; the phenomenon was one of hydrostatic pressure, 
conditioned by the contact relations of glass with mer- 

In the second experiment water was substituted for 


mercury; and, as glass is heavier than water, the glass 
block was replaced by a block of cork, to the bottom of 
which a thin plate of glass was cemented. When the 
cork was pushed down through the water and its glass 
face squeezed against the slab of glass at the bottom, it was 
found to adhere, but not permanently. The pressure ap- 
plied did not force all the water from between the glass ^ 
faces, and the remaining water film gradually thickened, 
until, in a few seconds or a few minutes, the cork was 
freed and rose to the surface. 

As regards conditions, the only essential difference be- 
tween the two experiments was in the liquids employed, 
and the properties of the liquids which determined the 
diverse results were the relations of internal to external 
molecular forces. Because the cohesive force of water is 
less strong than its force of adhesion to glass, water ' wets ' 
a glass surface; it is able to spread and interpose itself 
as a film between two glass surfaces pressed tightly to- 
gether; and when thus interposed it can not be forced 
out by pressure (at least, by such pressure as is involved 
in the problem of the tidal glacier). Because the cohe- 
sion of mercury is stronger than its adhesion to glass, it 
does not wet glass; and it does not tend to insinuate itself 
between closely approximated glass plates, but tends rather 
to withdraw from the interspace. 

These experiments indicate that the ability of the sea 
to penetrate, and communicate its pressure, along the 
contact surfaces of a tidal glacier and its bed, may depend 
on the cohesive force of water, as compared with its ad- 
hesive force in relation to ice and rock. It is a familiar 
fact that water wets both ice and rock; and the problem 
of the glacier is therefore better represented by the second 
of Day's experiments than by the first. 

Another factor of the problem should now be consid- 
ered, the temperature of the bottom ice and adjacent rock; 


for if these fall below the freezing temperature, a film of 
water can not exist between them. This is a subject 
which has received considerable attention, and although 
direct observation is impossible, there seems a good 
foundation for inference. By the aid of crevasses and 
borings it has been found that the upper ice (outside the 
neve region) has at all times a temperature of almost ex- 
actly 32°, seasonal and diurnal variation being confined 
to a very thin surface layer. The upper ice therefore has 
no cooling eflfect on the bottom ice. On the other hand, 
the bottom ice receives heat in three ways : Heat comes to 
it by conduction from the interior of the earth; heat is 
developed by the friction of ice and waste on the bed- 
rock; and the internal work of the flow of the ice devel- 
ops heat, of which a part is conducted downward. The 
bottom ice therefore maintains a temperature of 32*^ 
(more precisely, the freezing temperature corresponding 
to the pressure), and the adjacent rock is slightly warmer. 
There is a continual, though very slow, melting of the 
basal ice, and a film of water is maintained between it 
and the rock. It is probable that the streams of water 
which flow from glaciers all through the winter are sup- 
plied chiefly by basal melting; and we may further sup- 
pose that the tunnels through which those streams flow 
are connected with the thin water film by a graduated 
and ramifying system of minor passages. The ways 
which serve for the escape of the product of basal melting 
serve also, in the case of tidal glaciers, for the communi- 
cation of the hydrostatic pressure of the sea water. 

Statically considered, the film of water under the gla- 
cier is subject to a group of forces in equilibrium. The 
weight of the glacier presses on it and tends to expel it. 
This is resisted by the molecular forces associated with 
the contact faces of the film, and by the hydrostatic pres- 
sure of the sea water outside. As the film is added to by 


melting, and as the added water must escape by flow, 
there is also a dynamic factor, the viscous resistance to 
flow. These forces conjointly determine the thickness 
of the film. The film is thicker as the ice column is less, 
as the sea-water column is greater, and as the melting is 
more rapid. It is always very thin. 

We may now advantageously return to the question of 
the mode of support of the tidal glacier. Referring to the 
diagram (fig. 106), it is evident that the ice receives no 
support from the pressure of the sea , . , , ,.,, 

water against its frontal cliff*. It rests 
wholly on the film of water beneath 


it, and its pressure is communicated — — - ^^^^±LLj.U 

by the film, without loss, to the rock 'WMMMMMMMMMM 

beneath. The film is not a mere con- fig. 106. ideal sec- 

duit, communicating the static pres- ^'°^ °^ "^'^^^ glacier. 

^ r - 1 Showing relation of the 

sure of the sea water. If it were, the sea to the subgiaciai water- 

1 J , -I i J i_ film. The thickness of the 

ice would not be supported, because fii^^ is enormously exagger- 

that pressure is less than the down- ^^^^• 
ward pressure of the ice. In virtue of the molecular forces 
brought into play along the contact planes, the film has 
some of the properties of a solid. It is, in some sense, an 
elastic spring or cushion interposed between the ice and 
the rock. It performs its function of transmitting the 
pressure of the glacier to the rock bed quite indepen- 
dently of the presence of the sea. The pressure of the 
sea water modifies the infinitesimal thickness of the film, 
but does not prevent the rock bed from supporting (through 
the mediation of the film) the whole weight of the glacier. 
The last statement is subject to a single qualification. 
Wherever the feeble flow of water from basal melting 
maintains passages of more than capillary size, the molec- 
ular forces locally cease to dominate, and the hydrostatic 
pressure of the sea water contributes support to the 
glacier. To this extent, which must always be relatively 


very small, the pressure of a tidal glacier on its bed is 
diminished by the sea. 

It thus appears that there is no important difference, as 
respects pressure on the rock bed, between a glacier rest- 
ing on the land and one which is partly bathed by the 
water of a fiord; and, so far as glacial erosion is con- 
ditioned by pressure, the presence of the sea does not 
diminish the efficiency of the glacier. It is possible, how- 
ever, that the rate of erosion is affected by changes in the 
thickness of the capillary film of water. In the familiar 
case of the grindstone, the application of water modifies 
abrasion in two ways: It acts as a lubricant to diminish 
friction and reduce the efficiency of the stone ; and it acts 
as a carrier to remove the product of abrasion, prevent 
clogging, and thus enhance the efficiency. The sub- 
glacial film doubtless has both these functions, and both 
should be affected by its thickness; but as the two are 
antagonistic in relation to abrasion, it is not at once 
evident whether a thickening of the film should increase 
or diminish the erosive work of the glacier. 

It seems furthermore possible that the influence of film 
thickness on abrasion may not be the same for waste par- 
ticles of all sizes. It may be important with reference to 
the finest rock flour, where the diameters are of the same 
order of magnitude as the depth of the film, and quite un- 
important with reference to the work of sand and still 
coarser waste. And if we turn from abrasion to plucking, 
we seem to pass altogether out of the field of influence of 
the subglacial film. 

While it is not altogether easy to picture the combina- 
tion of molar and molecular forces associated with the sub- 
glacial film, and while it is still more difficult to analyze 
the effect of variation in that film on so complicated a proc- 
ess as glacial erosion, I am nevertheless confident that 
the influence of the sea in diminishing the pressure of a 


tidal glacier has been greatly overrated, and I discredit 
the supposed power of sea pressure to make important 
reduction of a glacier's efficiency for erosion. 

The preceding pages were submitted in manuscript to 
several friends competent to consider questions in molec- 
ular physics. Some of them think the ability of the sub- 
glacial film to resist expulsion has been overestimated, and 
especially that the film should not be assumed to exist 
between the bed-rock and the abrading angles of rock 
particles held in the ice. If it be true that abrading par- 
ticles are in absolute contact with the bed, and if it 
be further true that there is no film between the same 
particles and the partly enveloping ice, then parts of the 
glacier (regarded as a body of ice and fragmental rock) 
are directly supported by the bed. However important 
the distinction may be with reference to a complete 
theory of glacial abrasion, it seems to have little bearing 
on the question of pressure as here considered. We 
may conceive the whole glacier to consist of infinitesimal 
vertical columns, some terminating on the bed and sup- 
ported by it, others resting directly on a capillary film of 
water and thus indirectly supported by the bed, and yet 
others terminating on a water stratum of supercapillary 
depth. The last are sustained in part only by the hydro- 
static pressure communicated by the water stratum, and 
are otherwise upheld directly by their coherent neighbors 
of the first and second groups, and through them, indi- 
rectly, by the bed. 

It is a corollary to the general conclusion of this section 
that the existence of a fiord — that is, of a glacial trough 
partly occupied by an arm of the sea — is not demonstra- 
tive of a relatively low base-level at the time of its excava- 
tion. In the regions of the Alexander Archipelago and 
Prince William Sound there is independent reason to 
think the base-level was relatively low when the glaciers 


were largest, but the case is not strengthened by the pres- 
ence of the sea in the glacier channels. 


The resemblance which glaciers of the alpine type bear 
to streams of water, has impressed all observers, and it 
includes so many details of form and work that the phrase 
' river of ice ' seems more than a mere figure of speech. 
The fact that resemblances arrest attention, of course im- 
plies that there are also differences ; the contrast between 
the two materials, ice and water, is so extreme that cor- 
respondences in their behavior are unexpected and there- 
fore striking. If we disregard this fundamental difference 
in material and restrict attention to other causes controlling 
the phenomena, then we may say that some of the re- 
semblances between glaciers and rivers are of the nature 
of homologies, in that the causes of the corresponding 
features are like, and other resemblances are of the nature 
of analogies, in that the causes are unlike. Before attempt- 
ing a classification of resemblances on these lines, I shall 
enumerate the features of alpine glaciers which have been 
regarded, or which seem worthy of regard, as correspond- 
ing to similar features of rivers. 

Resemblances, — (i) Upland precipitation is gathered 
into streams which flow down the slopes. (2) As they 
flow they meet and join together, forming greater streams, 
which follow the main valleys or gorges. (3) Sometimes 
a stream parts against a prominence and reunites beyond 
it, thus surrounding an island (nunatak); (4) sometimes 
the parted members proceed independently, as distribu- 
taries. (5) All parts of the stream are subject to gain 
and loss of material. In an upper division gain is in 
excess; in a lower division, loss. In complete examples 
the maximum volume is in mid-course, and the stream 
ends distally by complete dissipation. (6) If the stream 


reaches the sea it is cut off, and its history of volume- 
change ends abruptly. 

The velocity of flow is greater (7) in the middle of the 
stream than at the sides, (8) at the top than at the bottom, 
(9) on the outer side of a bend than on the inner, (10) 
where the channel is narrow than where it is broad, (11) 
where the grade is high than where it is low, (12) in a 
large stream than in a small. 

(13) The surface of the stream, considered as to its 
major features, is smoother than the channel bed. (14) 
It is exceptionally high on the upstream side of an island, 
and exceptionally low on the downstream side. (15) It 
is roughened in detail by concealed prominences of the bot- 
tom, especially where the stream is shallow. (16) The 
surface descends normally in the direction of flow, but 
exceptionally, and for short distances, ascends. 

(17) The stream erodes its bed and the walls of its 
channel. (18) The chief tools of erosion are rock frag- 
ments carried by the stream. (19) The greater the 
velocity of the stream the more rapid the erosion (except, 
perhaps, in cirques). (20) The stream shapes its chan- 
nel, making the width several times greater than the 
depth. (21) The fully adjusted channel is broader in 
yielding material than in obdurate; but has uniform width 
in uniform material. (22) The contours of the adjusted 
channel are smooth curves of large radius. (23) The 
adjusted channels of large streams are broader and deeper 
than those of small streams. Where a large stream is 
joined by a small tributary and the surfaces of the two 
have the same level, the bed of the large stream lies lower 
than that of the small. (24) The erosion of the channel in 
gorges saps the cliff* walls, causing coarse waste to fall to 
the stream. (25) The waste from erosion and sapping is 
carried forward by the stream and eventually deposited. 

(26) Where a heavily loaded stream issues from a moun- 


tain gorge to an open valley, it sometimes deposits waste 
at the sides and beneath until it comes to flow in a walled 
causeway, or raised trough, of its own construction. It 
may then overflow a wall of the trough and assume a new 

Differences, — The features of difference are equally 
noteworthy. The speed of the glacier is very much 
slower than that of the river, being better expressed in 
feet per year than in feet per second. The rates of wax- 
ing and waning are correspondingly slow. A river flood 
is propagated downstream by the actual transfer of the 
water added about the upper course; a glacier flood is 
believed to be propagated downstream as a wave travel- 
ing more rapidly than the ice. The depth and width of a 
glacier are much larger, in relation to length, than those 
of a river. The threads of flow in a glacier run nearly 
parallel; in a river they weave freely in and out. That 
which falls to the back of a glacier, though much denser 
than the ice, does not sink to the bottom, but is carried for- 
ward as a back-load; only light materials float on a river. 
Most of the waste embedded in a glacier is moved along 
continuously; most of the waste constituting the load of 
a river is transported intermittently, being repeatedly 
picked up and laid down. 

Homologies and Analogies, — The gathering of ice into 
streams and its downward flow are caused by gravity, just 
as in the case of water. Most of the inequalities of veloc- 
ity are determined by gravity in conjunction with the fric- 
tion of the ice on the channel and the resistance of ice to 
internal shear; and the processes are essentially the same 
as with water. But the greater velocity on the outside 
of a bend involves an analogy only. The bending stream 
of ice distributes the velocities of its elements in such 

1 1. C. Russell. Eighth Ann. Rept. U. S. Geol. Survey, part i, pp. 337-342, 
360-366, 1889. 


way that the total work of channel friction and internal 
shear is a minimum, and that distribution throws the locus 
of maximum velocity to the outer side of the middle of the 
stream, but the general relations of the parts of the stream 
are not changed. The preservation of the general rela- 
tions is clearly shown by medial moraines, which map out 
surficial lines of flow. In the bending river, momentum 
is the controlling factor. Those fillets of the stream which 
above the bend moved fastest are thrown to the outer side 
of the curve, and the slower-moving fillets are crowded to 
the inner side. Usually the upper water moves to the 
outside of the bend and the lower water to the inside, so 
that there is a torsion of the body of water as a whole. 

The correspondences connected with gain and loss of 
material are close, and the parallelism can be traced 
through many details, but the processes are so different 
that the similarities of result can not be classed as homol- 
ogies. Alimentation of the glacier is primarily through 
snowfall, secondarily through rainfall, and there is storage 
in snow banks that subsequently descend as avalanches. 
Alimentation of the river is primarily through rainfall, 
secondarily through the melting of snow and ice, and 
there is storage in ground water that slowly issues in 
springs. Dissipation of the glacier is chiefly by melting 
and secondarily by evaporation. Dissipation of the river 
is wholly by evaporation, but part of its evaporation is 
preceded by absorption by the ground. 

The general evenness of the stream surface is deter- 
mined in each case by gravity, and so is its descent in the 
direction of flow; but the exceptional ascent in the direc- 
tion of flow has diverse causes. Where the passage of 
an ice stream over an embossment of its bed is expressed 
by a local rising of its surface, the imperfect adjustment 
of the surface is a result of viscosity; the upward slope of 
water under similar circumstances is a result of impetus 


or momentum. So, too, the roughening of the glacier 
surface where the flow is disturbed, by breaking into 
seracs and pinnacles, is analogous to, rather than homol- 
ogous with, the breaking of a river surface into waves. 
Viscosity causes the rupture of the ice, momentum the 
undulation of the water. The deflection of the viscous 
ice produces stresses and strains, some of which are ten- 
sile; in the depths of the stream the tensile stresses are 
balanced by compressive stresses due to the weight of 
overlying ice, but higher up the ice is overstrained and 
ruptured. When the swift-flowing water rises over an 
obstruction its momentum causes a portion to shoot above 
the normal level, and thus starts an undulation. 

The causes of the relation of channel depth to channel 
width are not sufficiently understood in the case of glaciers 
to warrant a comparison with rivers. The disparity of 
channel depth at the mouth of a small, fully-adjusted trib- 
utary is in each case a phenomenon of base-level con- 
trol. The surface level of the tributary is determined by 
the main stream, and if the tributary at any time erodes 
its channel rapidly, its stream becomes deeper, its velocity 
less, its power to erode less, and thus the tendency to 
deepen is limited. The resemblance of glaciers to rivers 
in this respect is a homology. 

The adjustment of channel contours to simple curves is 
brought about, in both classes of streams, by the more 
rapid erosion of projecting angles, but the work of the 
water is concentrated on these through the property of 
momentum, and the work of the ice through viscosity. 

The walled causeways sometimes built by streams of 
ice debouching onto a plain have a different process of 
construction from the similar causeways occasionally built 
by streams of water. The walls of the glacial causeway 
are made by the deposition of lateral moraines that had 
been carried chiefly as back-load. The walls of the fluvial 


causeway are made by the deposition of suspended waste 
through the slackening of spreading side currents. 

The salient fact brought out by these comparisons is 
that many features of rivers and river work which arise 
from inertia in association with swift motion are paralleled 
by features of glaciers and glacier work which arise from 
high viscosity in association with slow motion. In each 
kind of stream, changes in the direction of flow are caused 
by irregularities of the channel, and complex series of 
phenomena arise from the resistance of the current to 
deflection. These series are strikingly parallel, but the 
resistance to deflection is occasioned in one case by 
momentum and in the other by viscosity. 


Abercrombie, Captain W. R., map 72 
Acknowledgments, assistance by col- 
leagues ID 
photographic material 5 
Alexander Archipelago, depth of fiords 


fiords 150-159 

pre-glacial topography 130 
Amherst Glacier 82, 82 (pi.) 
Annette Island 130 

Baranof Cape 132, 132 (pi.) 
Barry Glacier 90-93 
Base-level, pre-glacial 134 
Base-level plains. See Peneplains. 
Base-levels, Pleistocene 162-166, 172, 

Becker, G. F., cited on glacial erosion 
cited on statics of tidal glacier 212 
Behm Canal 128, 134, 140, 159 
Bendeleben Mountains 193 
Bering Sea 186-194 
Berner Bay 127, 140 
Biological Survey, U. S., acknowledg- 
ments V 
Alaska photographs 5, 6 
photograph of Welles ley Glacier 

photograph of Yale Glacier 83 
Bonney, T. G., cited on hanging val- 
leys 114 
Boundary Commission, ^te Canadian 
International Boundary Commission. 
Brabazon, A. J., photographs 27, 28, 29, 


Brady Glacier 45, 106, 123 

British Columbia, glaciation of coastal 

region 141-150 
Brooks, A. H., geology of Seward 

Peninsula 192 
Bryn Mawr Glacier, frontispiece, 87, 

88, 175 

Canadian International Boundary 
Commission, acknowledgments v 
maps and photographs about Gla- 
cier Bay 16, 25, 26, 28, 29, 38 
maps and photographs about Lynn 

Canal 12, 13, 15 
maps and photographs about Ya- 

kutat Bay 46, 62, 65, 67, 209 
photograph series 6, 122 
photographs of fiords 158 
photographs of upland topography 
123, 125, 127, 128, 140 
Cape Baranof 132, 132 (pi.) 
Cape Spencer peneplain 123-126, 141 
Cataract Glacier 94 
Cascading glacier, Nunatak Fiord 61 

glaciers, College Fiord 89, 175 
Caves in Barry Glacier 91 
Chamberlin, T. C, cited on hanging 

valleys 114 
Charpentier Glacier 34-37 
Chatham Strait 134, 1 51-155 
Chatham Strait Glacier 162 
Chilkat Inlet 13, 15 
Chilkoot Lake 141, 154 
Chiniak Bay 179 
Cirques 141 
Clarence Strait 134 



Coast Survey, U. S., maps 12, 15, 26, 

49. 50. 1 19 

photographs 122, 187 

soundings 158 
College Fiord, existing glaciers 81-89 

map 80 (pi.) 

Pleistocene glaciation 175-176 
Collier, A. J., cited on geology of 

Seward Peninsula 192 
Columbia Bay 73, 81, 174 
Columbia Glacier 70 (pl.)» 71-81, 104 

surface features 200 
Columbia River 137 
Coville, F. v., acknowledgments 10 

observation of young spruces 78 

visit to Columbia Glacier 71 
Crescent Glacier 82, 82 (pi.) 
Crevasse cycle 198-203 
Curtis, E. S., official series of views 5 

photographs of glaciers 6, 74 (pl.)> 

86, 88, 94, 94 (pi.), 95 
photographs of Wrangell and Sitka 
132 (pi.) 
Cushing, H. P., geologic studies in 
Glacier Bay 16, 17 

Dall, W. H., acknowledgments 10 
cited on Kadiak terraces 179 
cited on Pleistocene glaciation 187 
observations of Grewingk Glacier 

98, 100 
Pleistocene shells 165 
Dana, J. D., cited on glacial erosion by 

plucking 207 
Davidson Glacier 12-16, 105, 157 

compared with glacier of Russell 

Fiord 51 
compared with Pleistocene glaciers 
of Fairweather Range 169 
Davis, W. M., cited on hanging val- 
leys 114 
Dawson, G. M., Alexander Archipelago 
cited on ancient glaciers of Van- 
couver Island region 145 
cited on geology of coast ranges 
Day, A. L., experiments 212 
Denudation. See Erosion. 

Discovery Passage 143 
Disenchantment Bay 46-49, 62 (pi.), 

63-70, 104 
Doran Strait 90 

Disenchantment Bay Glacier 49 
Douglas Island 132, 165 
Drift 35, 161 

influence on topography of glaciers 
Dutch Harbor 185, 210 

Earthquake 23 

Emerson, B. K., acknowledgments 10 
Erosion, by glaciers and rivers com- 
pared 218-223 

by Pleistocene glaciers 139-163 

by tidal glaciers 210-217 

fiord 162 

principles of glacial 203-210 
Estuaries, coast of Washington 137 

Fairweather Range 39, 123, 125, 126 
existing glaciers 148-150 
La Perouse Glacier 39-45 
Pleistocene glaciation 166-170 
variations of glaciers 105, 107 

Fiord erosion 162 

Fiords, Inside Passage 141-150 
strike 125, 142-157 
transverse 142, 157-159 

Fiords and islands 1 19-122 

Fish Commission, U. S., photographs 
44i 65, 67 

Fort Wrangell 132 (pi.) 

Eraser Reach, hanging valleys 145-147 

Gannett, Henry, acknowledgments 10 
cited on cascading glaciers 175 
cited on hanging valleys 114 
map of Alaska 1 (pi.) 
map of Disenchantment Bay 

62 (pi.), 65 
map of Hidden Glacier 52 (pi.) 
map of Muir Inlet 21 
map of Nunatak Glacier 58 (pi.), 

map of Port Wells 80 (pi. ), 81 
route 4 
Garwood, E. J., cited on hanging val- 
leys 114 



Gastineau Channel 132, 165 
Geikie Glacier 38 
Geikie Inlet 38 

Geological Survey, U. S., acknowledg- 
ments V 

photographs bj 5, 6 
Georgia, Gulf of 134 
Glacial deposits i6i 
Glaciated rocks 206 (pi.) 
Glaciation, Pleistocene 113-194 
Glacier, Amherst 82, 82 (pi.) 

Barry 90-93 

Brady 45, 106, 123 

Bryn Mawr frontispiece, 87, 88, 

Cataract 94 
Charpentier 34-37 
Chatham Strait 162 
Columbia 70 (pi.), 71-81, 104, 200 
Crescent 82, 82 (pi.) 
Davidson 12-16, 51, 105, 157, 169 
Disenchantment Bay 49 
Geikie 38 

Grand Pacific 26-33 
Grewingk 97-102, 104 
Harriman 94, 94 (pi.) 
Harvard 82, 84 

Hidden 47, 52 (pi.), 52-58, 202 
Hubbard 47-49, 62 (pi.), 63-70, 

Hugh Miller 34, 35-38 
Johns Hopkins 27, 28, 32, 106 
La Perouse 39-45, 104, 167 
Malaspina 46, 49, 106 
Muir 20-25, 198. 199 
Nunatak 47, 58 (pi.), 58-63 
Puget Sound 135, 139 
Queen Charlotte Sound 145 
Radcliffe, 84-87, 175 
Reid 27, 28-32 
Roaring 96 
Serpentine 93 
Smith 86, 87, 175 
Strait of Georgia 145 
Surprise 94 
Turner 47, 62 (pi.), 64, 66 (pi.), 

Vassar 88, 175 
Wellesley 175 

Glacier, Wood 38 

Yale 82, 83 
Glacier Bay 16-39 

variations of glaciers 103 
Glaciers, annual cycle 22 

basal melting 214 

cascading 61, 89, 175 

causes of variations 102-112 

classification 10 

College Fiord 81-89, ^75-17^ 

comparison with rivers 218-223 

dates of photographs 5 

existing 7-112 

general distribution 7-9 

Glacier Bay 20-39 

hanging 11, 114-119 

Lynn Canal 11-16, 126, 134, 150- 

piedmont 10 
Pleistocene 1 13-194 
Port Wells 81 
pressure and erosive power in 

fiords 210-217 
principles of erosion and sculpture 

relation to forests 41-44, 45, 49, 51, 
52, 76, 78-80, 88, 92, 95, 96, 98, 
loi, 106 
sculpture by Pleistocene 139-163 
summary of modern changes I03- 

surface characters 196-203 
tidal II, 210-218 
Glenn and Abercrombie, map 72 
Graham Island 165 
Grand Pacific Glacier 26-33 
Gray Harbor 137 
Grenville Channel 146 
Grewingk Glacier 97-102, 104 
Gulf coast, summary of Pleistocene 

geology 182-185 
Gulf of Georgia 134 

Haenke Island, contact with Hubbard 
Glacier 48, 70 

photographs 65, 67 

position 47, 50 
Hall Island 188 
Hanging glaciers 11, 114-119 



Hanging glaciers, in Harriman Fiord 

93» 94. 95. 176. 

See also Hanging Valleys. 
Hanging valley, Kadiak Island 181 

Plover Bay 191 

Princess Royal Island 116 (pi.) 
Hanging valleys 114-119 

Inside Passage 142-150 

Lynn Canal 153, 154 

Nunatak Fiord 60, 61 

Yakutat Bay 170 
Harriman, E. H., photograph by 95 
Harriman Fiord 80 (pi.). 89-97 

Pleistocene glaciation 176 
Harriman Glacier 94, 94 (pi.) 
Harvard Glacier 82, 84 
Hayes, C. W., cited on height of n^vd 

line 8 
Hecate Strait 134 

Hidden Glacier 47, 52-58, 52 (pi.), 202 
High mountain district 166-173 
Hinchinbrook Island 174 
Hubbard, Gardiner G. 63 
Hubbard Glacier 47-49, 62 (pi.), 63- 

70, 106 
Hugh Miller Glacier 34, 35-38 
Hugh Miller Inlet 34-38 

Ice. See Glacier and Viscosity. 
Iceberg waves 3 1 , 69 
Icebergs, Muir Inlet 24 

Reid Inlet 33 
Iliuliuk 186, 210 

Inertia of rivers compared with vis- 
cosity of glaciers 223 
Inland Passage district 1 19-166 
Inside Passage 121, 142-150 
Inverarity, D. G., photographs by 62, 

92, 94 (pi.) 
Island, Annette 130 

Douglas 132, 165 

Graham 165 

Haenke 47, 48, 50, 65, 67, 70 

Hall 188 

Hinchinbrook 174 

Kadiak 177-182, 189 

Montague 176 

Osier 47, 65, 69 

Pitt 146 

Island, Princess Royal 116 (pi.), 145, 


St. Matthew 188, 193 

St. Paul 187 

Unalaska 185 

Vancouver 143-146, 165 
Islands, Queen Charlotte 165 
Islands and fiords 1 19-122 
Itinerary 3 

Johns Hopkins Glacier 27, 28, 32, 106 
Johnstone Strait 143 

Kadiak Island 177-182, 189 

Kenai Peninsula 97, 177 

Kerr, M. B. i 

Kettle-holes, near Grewingk Glacier 

near Hidden Glacier 54 (pi.), 54- 

Kigluaik Mountains 192 
Klotz, Otto J., cited on variations of 
glaciers 45, 103 

La Perouse, J. F. de G., exploration of 

Lituya Bay 45 
observations of Disenchantment 

Bay Glacier 51 
La Perouse Glacier 39-45, 104, 167 
La Perouse Peak 123 
Lituya Bay 149, 150, 167 
Lowe Inlet 148 
Lynn Canal, configuration of bottom 

glaciers 11-16 

origin of 150-157 

upland topography near 126 

McArthur, J. J., photograph by 13 
Malaspina, A., visit to Disenchant- 
ment Bay 48 
Malaspina Glacier 46, 49, 106 
Map of Alaska, showing route i (pi.) 
Mendenhall, W. C, cited on glaciation 

of Alaska Peninsula 192 
Merriam, C. Hart, photographs by 5, 6 
photograph of Amherst Glacier 

82, 82 (pi.) 
preface v 
Montague Island 176 
Moraine, section 56 (pi.) 



Moraine-building at sea front 163-165, 

Moraine delta, Davidson Glacier 13-16 

Grewingk Glacier 97 

Hidden Glacier 52-57 

Muir Glacier 19 
Moraines, base of Fairweather Range 

near Yakutat Bay 171 

overridden by Columbia Glacier 


under Yakutat Bay 49-51 
Moraines, push-, Columbia Glacier 

Grewingk Glacier 100 

La Perouse Glacier 42 
Muir, John, acknowledgments 10 

cited on Hugh Miller Glacier 35 

cited on Pleistocene glaciation of 
Bering Sea region 186 

exploration of Glacier Bay 16, 17 

route in 1899 3, 4 

studies of Brady Glacier 45 

studies of Geikie Inlet 38 

studies of Grand Pacific Glacier 
26, 29, 32 

studies of Muir Glacier 21 

visit to Harriman Fiord 81 
Muir Glacier 20-25 

limit of crevasses 198 

seracs 199 
Muir Inlet, glacial erosion 207 

ice pack 24 

map 21 

Nanaimo 165 

Nordenskiold, A. E., cited on glacia- 
tion of Siberia and Alaska 187, 192 
Nunatak Fiord 46, 58, 61, 62 
Nunatak Glacier 47, 58 (pi.) > 58-63 

Ocean Cape 170 

Ogilvie W., photograph of Davidson 

Glacier 12 
Orca 174 
Osier Island 47 

ice-fall waves 69 

photographs 65 

Palache, Charles, acknowledgments 10 

Palache, at Columbia Glacier 71 

route 3, 4 
Penck, A., cited on hanging valleys 

Peneplains, high, Kadiak Island 177, 

Kenai Peninsula 177, 183 

Prince William Sound 172, 183 

southeastern Alaska 123-130, 183 
Peneplains, low, Alexander Archi- 
pelago 130-134, 159, 183 

glaciation of 209 

Kadiak Island 179, 180, 182, 183 
Photographs of glaciers 4 
Pitt Island 146 

Pleistocene glaciation 1 13-194 
Plover Bay 190 
Plucking 203, 205, 207-209 
Port Clarence 188 
Portland Canal 134, 159 
Port Wells 80 (pi.) 

existing glaciers 81-97, 104 

Pleistocene glaciation 175 
Pre-Pleistocene topography 122-139 
Pressure of tidal glaciers 210-217 
Prince William Sound, ancient base- 
levels 172 

existing glaciers 71-97 

map 80 (pi.) 

Pleistocene glaciation 173-176 
Princess Royal Island, hanging valleys 

116 (pi.), 145, 146 
Puget Sound 135 
Puget Sound Glacier 135, 139 
Push-moraines, Columbia Glacier 76- 

Grewingk Glacier 100 

La Perouse Glacier 42 

Queen Charlotte Islands 165 
Queen Charlotte Sound 134 
Queen Charlotte Sound Glacier 145 

Rabot, Charles, cited on cause of gla- 
cial variation 106 
Radcliffe Glacier 84-87, 175 
Reid, H. F., cited on cause of glacial 
variation 109 
cited on variations of glaciers 103 
nomenclature of glaciers 11 



Reid, H. F., studies of Muir Glacier 
21, 23 

survey of Glacier Bay i, 16, 17, 
25, 26, 28, 29, 38 

Reid Glacier 27, 28-32 

Reid Inlet 25-34 

Rivers of ice and water compared 218- 

Roaring Glacier 96 

Rock table 201 

Route, map of i (pi.) 
of author 3 
of Expedition i, 2, 3 

Russell, I. C, cited on buried forest 52 
cited on Disenchantment Bay and 

Russell Fiord 48 
cited on glaciation of Unalaska 

Island 185 
cited on Hidden Glacier 56 
cited on marine shells 173 
cited on nomenclature 1 1 
cited on Nunatak Glacier 62 
cited on variations of glaciers 103 
classification of glaciers 10 
Davidson Glacier 13 
photograph of Hubbard Glacier 65 
photograph of Turner Glacier 66, 

66 (pi.) 
studies in Glacier Bay 16, 17 
survey of Malaspina Glacier i, 46 

Russell Fiord 46, 47, 51-63, 69 
types of glacial erosion 208 

St. Elias Alps 166 

height of n6v6 line 8 

recent uplift 173 
St. Matthew Island 188, 193 
St. Paul Island 187 
Sea-levels, Pleistocene 162-166, 172 
Seracs 198 

Serpentine Glacier 93 
Seward Peninsula 192 
Shaler, N. S., cited on pressure melt- 
ing 204 
Shells, Pleistocene 165 
Siberia 190, 193 
Sitka 131, 132 (pi.) 
Smith Glacier 86, 87, 175 
Speel River 158 

Spencer Cape, peneplain 123-126, 141 

Spurr, J. E., cited on terraces of Un- 
alaska Island 186 

Stevens Passage 134 

Strait of Georgia 134 

Strait of Georgia Glacier 145 

Streams of ice and water compared 

Surprise Glacier 94 

Terrace, Annette Island 130 

near La Perouse Glacier 167 

Seward Peninsula 193 

Spruce Island 179 
Terraces, Prince William Sound 176 

Unalaska Bay 186 

Yakutat Bay 170 
Tidal glaciers 1 1 
Tidal glaciers competent to erode 210- 

Tittmann, O. H., ice pack in Muir In- 
let 24 
Tracy Arm 159 

Turner Glacier 47, 62 (pi.), 64, 66 
(pi.), 66-69 

Unalaska Bay, glaciation of 209 

Unalaska Island 185 

U. S. Biological Survey, acknowledg- 
ments iii 
Alaska photographs 5, 6 
photograph of Wellesley Glacier 88 
photograph of Yale Glacier 83 

U. S. Coast Survey, maps 12, 15, 26, 

49. 50» 119 
photographs 122, 187 
soundings 158 
U. S. Fish Commission, photographs 

44. 65. 67 
U. S. Geological Survey, acknowledg- 
ments V 
photographs by 5, 6 
Uyak Bay 178 

Vancouver, George, cited on Columbia 

Glacier 71 
map of Columbia Bay 71, 80 
observations in Disenchantment 

Bay 48 
observations in Glacier Bay 17 



Vancouver, George, observation of 

Brady Glacier 45 
Vancouver Island, hanging vallejs 

marine clays 165 
Variations of glaciers 102-112 
Vassar Glacier 88, 175 
Viscosity of ice, compared with inertia 
of rivers 223 

in relation to glacial erosion 204 

Walker Bay 128 

Waste plain, Davidson Glacier, 13- 

Grewingk Glacier 97 

Hidden Glacier 52-57 
Wellesley Glacier 175 - 

Whidby, Joseph, cited on Columbia 

Glacier 71 
Willapa Bay 137 
Willis, Bailey, cited on Puget Sound 

Glacier 135 
Wood Glacier 38 
Wrangell 132 (pi.) 

Wright, G. F., studies in Glacier Bay 
16, 17, 21 

Yakutat Bay, submerged moraines 49- 

terraces 170 
Yukutat Bay region 45-70 
Yakutat village 170 
Yale Glacier 82, 83 
York Mountains 193 



Harriman Alaska Ez 
Harriman Alaska'