To search for more bones, we visually inspect regions (|\(l\)|<30\(^{\circ}\), |\(b\)|<1\(^{\circ}\)) where arms are predicted to lie according to our current understanding of the Milky Way’s structure; the expected (l,b,v) paths of the Galactic arms are calculated using a log-spiral approximation, as described in the literature. The predicted positions of the Galactic arms (Scutum-Centaurus, Carina-Sagittarius, Norma-Cygnus, and Perseus) are overlaid on three-color GLIMPSE Spitzer \citep{Benjamin_2003,Churchwell_2009} images in World Wide Telescope—a tool that facilitates easy visualization of several layers of data at scales from the full sky down to the highest-resolution details. As part of our initial criteria, we search for long, largely continuous, filamentary mid-infrared extinction features that are near and roughly parallel to the Galactic mid-plane. This initial inspection yielded about 15 Bone candidates, and a video showing how this search worked in WWT is available at tinyurl.com/morenessies.
Regardless of this initial visual inspection, the true nature of these filaments, and their association with a spiral feature, can only be established by looking at radial velocity data. The filament must have similar line-of-sight velocities along its length. Moreover, the measured radial velocities should be very close to those predicted by the Milky Way’s rotation curve for arms at a known distance. To investigate the velocity structure of these fifteen filaments, we employ radial velocity data from four separate radio surveys: HOPS \citep{Purcell_2012,Walsh_2011}, MALT90 \citep{Foster_2011,Jackson_2013}, BGPS \citep{Schlingman_2011} and GRS \citep{Jackson_2006}. The HOPS, MALT90, and BGPS surveys are all geared towards probing dense regions hosting the early stages of high mass star formation. From the HOPS survey, we utilize the thermal line from ammonia. With a critical density of about \(10^{4}\textrm{ cm}^{-3}\), ammonia traces dense molecular gas and is often found in dense, cool clouds with temperatures less than 100 K \citep{Purcell_2012}. The \(\mathrm{N_2H^{+}}\) and \(\textrm{HCO}^{+}\) thermal line we utilize from the MALT90 and BGPS surveys are also particularly strong in cold dense regions. While the HOPS and BGPS surveys are complete over 100 and 170 square degrees, respectively, MALT90 was a follow-up survey targeted towards \(\approx2000\) dense molecular clumps first identified in the ATLASGAL 870 \(\mu\textrm{m}\) Galactic plane survey \citep{Schuller_2009}. As infrared dark clouds tend to harbor cool, high density clumps of gas which fuel the formation of massive stars, all three of these databases contain spectra for hundreds of regions within the longitude range of the potential bone-like filaments.
In cases where HOPS, MALT90, and BGPS catalog data are not available along the extinction feature, we were also able to extract spectra from GRS (high resolution \(^{13}\)CO (1-0) data) and MALT90 p-p-v cubes using the spectrum extracter tool in Glue. A demonstration of the procedure used to extract velocities in Glue is shown in figure \ref{fig:glue}. As CO traces lower density gas (on average \(10^2 \textrm{ cm}^{-3}\)) and \(\mathrm{N_2H+}\), \(\textrm{HCO}^+\), and \(\textrm{NH}_3\) trace high density gas (\(>10^4 \textrm{ cm}^{-3}\)), the dense gas sources provide much stronger evidence for the velocity of cold, dense, filamentary IRDCs. When dense gas sources were not available, the complete and unbiased high resolution GRS survey, although less desirable, allows us to roughly gauge the velocity along entire lengths of filaments. In filaments composed entirely of GRS spectra, we took HOPS spectra over the entire filament using Glue and confirmed that this HOPS-determined velocity agreed with GRS-determined average velocity to within 5 km/s.
By overlaying the HOPS, MALT90, BGPS, and GRS determined velocities on a p-v diagram of CO emission, we determine whether these filaments are physical spines or simply a chance projection of mid-infrared extinction features along our line-of-sight. For this study, we use the whole-galaxy \citet{Dame_2001} CO survey to locate each of the arms in p-p-v space. Of the approximately fifteen candidates identified visually, ten of these candidates are within 10 km/s of the Scutum-Centaurus and Norma-Cygnus arms. The central coordinates for these ten filaments, along with their average lengths, LSR velocities, and distances, are listed in figure \ref{fig:candidates}. We plot these ten candidates in p-p-v space, as shown in figure \ref{fig:skeleton}. In addition to showing our Bone candidates, we show several different predictions of the positions of two spiral arms toward the inner Galaxy in longitude-velocity space, from \citet{Dame_2011}, \citet{Sanna_2014}, \citet{Shane_1972}, and \citet{Vallee_2008}. For reference, we note that the new BeSSeL (maser) results from \citet{Sato_2014} in the first quadrant favor the oldest, HI-based \citet{Shane_1972}, fits for the Scutum arm.
After narrowing down our list to ten filaments with kinematic structure consistent with Galactic rotation, we develop a set of quantitative criteria for objects to be called “bones:”
Largely continuous mid-infrared extinction feature
Roughly parallel to the Galactic plane
Within 20 pc of the physical Galactic mid-plane
Within 10 km/s of the global-log spiral fit to any Milky Way arm
No abrupt shifts in velocity (of more than 3 km/s per 10 pc) within extinction feature
Projected aspect ratio \(\ge 50:1\)
We calculate the aspect ratios and masses per unit length for all ten filaments, along with other parameters, which are summarized in figure \ref{fig:mass_of_bones}. Of the ten filaments with velocities consistent with galactic rotation, six of these meet all six bone criteria: candidates 1, 3, 5, 7, 9, and 10. However, it is important to note that some of the above criteria will likely be modified in the long run, as we learn more about the Skeleton of the Milky Way. Given our limited a priori knowledge of the Galaxy’s structure, it is presently easier to find Bones that are spine-like, lying along arms with velocities predicted by extant modeling (criteria 1, 5), and harder to find spurs off those arms or inter-arm features, the velocities of which are hard to predict well. Similarly, criterion 6 does not allow for projection effects in imposing an aspect ratio limit. As we learn more about spiral structure from simulations and modeling, these criteria will also be adjusted to allow for Bone-like features that represent spurs, inter-arm structures, and/or foreshortened structures lying close to our line of sight.