Referee 1's review:

This paper is unacceptable for a variety of reasons and has so many fatal flaws that there are no reasons to ask for revisions. When I accepted the invitation for review, I anticipated a well-reasoned, well-documented manuscript, but this paper is the opposite. I list only a few of the problems, but it is full of citation errors, unreliable correlations, statistical manipulations and lacks essential documentation.

I start with documentation. There is no list of the skeletal elements they are describing nor their context. Some of the specimens were surface collected, others were excavated, but there are no associated skeletal list. They state that "exploitation of these caves in 2006 revealed substantial numbers of fragmentary and complete human remains," but as far as I can tell the only complete remains were three tali, one of which is battered. If there are "numbers of complete human remains" they need to list them. But, minimally, to determine the importance of this material, there has to be some kind of inventory. The problems extend to dating. I attempted to figure out which specimens were radiometrically dated, but could only determine that none of the specimens in their tables are directly dated. So, given that some (all?) material is in the cave due to wave action along with a number being surface finds, there is no context for the bones and especially no association of the elements with each other.

Statistical manipulations and false correlations: Given the above, it is nonsensical to make ratios of craniofacial, mandibular or dental dimensions to postcranial dimensions. This is unacceptable and, despite their attempt to use the smallest/smallest, largest/largest manipulation, there is no intuitive reason nor cited study to show this technique is even remotely reliable. In this regard, it is hard to understand their comment on p. 9: "Due to the present lack of articulated specimens, we are unable to comment on hand and foot size relative to body size." Why is do they make ratios for relative face size, but not relative foot size? There are also problems with the way data are presented. I did not check a lot, but have no reason to suspect there are not more problems with data evaluation. They maintain on page 7 that "The symphyseal region tends to be vertically oriented, and the mean symphyseal angle [24] (92.5{degree sign} {plus minus} 3.1{degree sign}, n=4) is closer to that of archaic humans (90{degree sign} {plus minus} 5.1{degree sign}, n=10) than that of recent humans (range of means: 98.5{degree sign} - 106.1{degree sign}) [23]." Despite the fact that the comparative samples are really small and there is no statistical test for this assertion, the comparison to mean values here is misleading. According to the table they cite in 24 (not 23 which is about Japanese long bones), the San range is 92-114; the Zulu 87-125 and the Tolai 88-110. The fossil sample range is 95-121. The Palau chins are totally within the ranges all groups and the modern and fossil samples overlap extensively. As for body size, there is almost zero reliability in going from an acetabular maximum diameter to a body mass to a stature estimate. They attempt no other way to estimate body height based on the Palau remains and I seriously doubt these estimates are correct. They are certainly more likely to be in error, given the indirect technique. Similar objections concern their efforts to estimate cranial capacity from facial dimensions. As far as I know there has never been an attempt to do this before now and there is no reason to accept it is even remotely accurate, despite their cursory presentation in S1. Parenthetically, if one substituted in the Liang Bua 1 facial dimensions, they would not get anything close to 400 cm3. Finally, the following sentence (on p. 7) is another example of their faulty statistical reasoning: "The femoral head diameter (35.2 mm) of this specimen falls nearly one standard deviation below the reported mean for Andamanese pygmies [22]." They make a similar assertion on p.5 about tibial bicondylar breadth in Japanese females. Since when is falling 'nearly one standard deviation' below anything of importance and Japanese females are not pygmies? And, I have specimens in my samples of where maximum femur diameter is as small as 35.2mm and these data are not drawn from pygmies nor micropygmies (like LB1).

Citation errors: They cite Richards 2006 in the following: "Pygmy populations are known from mainland tropical forests and tropical island settings in Africa and Southeast Asia, reflecting parallel cases of dwarfing in response to the combined factors of relative genetic isolation, a reduced resource base, hot and humid climates, and (in certain island contexts) an absence of terrestrial predators [1]" on page 1-2. Not Richards, nor anyone else, has maintained that an "absence of terrestrial predators" is related to human dwarfing and I remind the authors that the 'hobbit' is associated with the largest living reptiles. There are also mis-citations of the 'microlithic and laminar stone tools" (p. 2), which are now are more closely linked to Olduwan tools and in several places statements about LB1. In one place on P2 they say that the Liang Bua fossils are characterized by the 'absence of projecting chins' then nine below describe a shared character between Palau and LB as "a weakly developed mental eminence." Also, how can 'dysplasias' (p. 2) be a taxonomic trait?

In short, this paper is so full of errors and misinterpretations that it is completely unacceptable. At best they have discovered on Palau some small individuals, but this is not well documented by them and not especially important since they are also found on Flores and other SE Asian islands. It adds little but confusion to the issue of the taxonomic postion of the Liang Bua material.

* * * * * * * * * *
N.B. These are the comments made by the referee when reviewing an earlier version of this paper. Prior to publication the manuscript has been revised in light of these comments and to address other editorial requirements.