Notices of New BooTcs. 409 Latitude. As an instance of inaccuracy we may state, that no attention seems to have been paid to Mr Ellis's arguments for a road over Mont Cenis : indeed the Alps are represented (PI. 7.) as discontinuous in that very spot, though the pass is nearly 7000 feet high. But there are other and graver faults to be noticed which this Atlas shares more or less with most other English Atlases. Modern names, the natural landmarks of our memory, are (in classical maps) either altogether omitted or irregu- larly inserted: in the present case, confined (except in the map of Britannia and a few mountains and why these exceptions ?) to the rivers. Why is no attempt made to introduce more physical geography (if it may be so expressed) into our General Atlases ? For instance, the comparative heights of table-lands are much more important than those of the moun- tain-chains, and yet are always neglected. Might not different shades of colour be used to denote these, and dotted lines alone serve for political boundaries ? The mountain-chains themselves are almost always care- lessly shaded. To take a common instance. Mount Athos (see PI. 14), which rises directly from the sea to the height of above 6000 feet, is not nearly so deeply shaded as the Cumberland hills (PI. 4), though the former map is on a scale of 32, the latter of 41, miles to the inch. It would be a great addition to our Atlases if vertical sections were intro- duced, like those excellent ones in Grimm Mahlmann and Kiepert's Atlas of Asia. Again, why are not the usual courses of navigation marked just as much as roads on land ? Why, except to confuse the eye, are the square borders, &c. of the maps retained (PI. 10, 15), when the parallels are all at an angle with them ? Why is not the proportion the scale bears to natural size, or the number of miles to the inch, stated on each map ? And above all, why are not the number of scales reduced, and those retained made some easy multiples of each other ? Mr Hughes himself, in his "Mathematical Geography," has admitted and blamed this huge defect, and yet in this Atlas (to omit the map of the World, the small plans of towns, and very slight differences), in 23 maps there are 15 different scales, and of these, only 2 common to 3 maps ! It is only fair to Mr Johnston to say that in his Classical Atlas, of the last two defects the first is entirely, and the second partially, supplied.] H. J. R. Etymologisch.es Worterbuch der romanischen Sprachen von Friedeich Diez ; Bonn, 1853, pp. 782. [This learned and elaborate glossary (for dictionary we can hardly call it) is sure to find a very cordial welcome in the library of most philologers. It consists of two parts : the former is devoted to a scientific analysis of words that kept their place in all, or nearly all, the principal languages whose parent-stock is Latin (Italian, Proven9al, French and Spanish) ; while the latter proceeds to examine the more difficult words now extant in some one or other of those branches of the Romanic family. The author's intimate acquaintance with the German language, both in its earlier and later form, enables him to point
