There is a world somewhere between reality and fiction. Although ignored by many, it is very real and so are those living in it. This forum is about the natural world. Here, wild animals will be heard and respected. The forum offers a glimpse into an unknown world as well as a room with a view on the present and the future. Anyone able to speak on behalf of those living in the emerald forest and the deep blue sea is invited to join.
Mandible length of fragment is affected by allometry. The "mandible length" measurement in the tiger paper is not the same as complete mandible length. Rather, you will see that as the mandible size increased, the surrogate "mandible length" comprises proportionately less and less of the total mandible. In other words, positive allometry. Here is a visual:
*This image is copyright of its original author
The larger specimens have a shorter horizontal ramus and surrogate mandible length compared to total mandible length. You will see here that comparison of the surrogate "mandible length" between 1a and 3a will underestimate 3a by a massive 17% (65/76). Therefore, the clean isometric relation of total mandible length does not apply. This is why height and width dimensions of the ramus are important, as there is less allometric effect on estimation theoretically when we pool together many measurements. This is also probably why multiple measurements are used in the estimate, to make up for this error. It is important to know these allometric relationships before choosing a single measurement as "most accurate" as clearly, the surrogate mandible length has hefty underestimation.
The comparative specimen used to estimate via mandible length is very small. As such, using a scale factor of 3 would underestimate the mass as the actual scale factor is greater than 3.
It may happen in time that the Christiansen estimations show to be underestimates across the board. The database is measured from digital photos, which can result in measurements a bit exaggerated as compared to in person measurements. This underestimates any specimen not also measured digitally.