\input{preface} \chapter{Conclusions} \label{ch:conclusion} The key conclusions of this thesis are divided into two components. The first centers on the Bolocam Galactic Plane Survey and its high-level results. The BGPS has laid the grounds for an extensive study of dense gas within our Galaxy. It is the first blind survey of the Galactic plane at millimeter wavelengths, where optically thin dust emission dominates the observed signal. \begin{itemize} \item The BGPS has had two data releases and its pipeline has been well-characterized. The angular transfer function drops from $100\%$ recovery at 100\arcsec to $\sim50\%$ recovery at 300\arcsec. \item The BGPS map power spectra, when compared with Herschel Hi-Gal power spectra, indicate that in some portions of the galaxy, the smallest scales are \emph{warmer} than the largest scales, hinting that internal heating by forming young stars is significant. \item There are 3 massive proto-clusters in the northern Galactic plane, G10.62, W49, and W51. \item There are about 20 `clumps' of mass $M\gtrsim10^4$ \msun in the northern plane. \item All of these clumps are forming massive stars at present, implying that the starless timescale for the parent clumps is $\tau_{starless}<0.5$ Myr. \item The BGPS and comparable ground-based surveys are excellent tools for identifying the precursors to massive clusters. Because the galaxy is transparent at 1.1 mm, the BGPS can be used for galaxy-wide population analyses \item Careful distance determination is crucial for population studies \item There is substantial free-free `contamination' in the brightest BGPS sources, but these sources are nonetheless dust rich. \end{itemize} % I examined the brightest sources within the BGPS, discovering 18 with masses % $M>10^4$ \msun, large enough to form bound clusters. These sources are all % actively star-forming and can now be used for unbiased proto-cluster population % studies. These observations allow us to place an upper limit on the starless % lifetimes of young massive clusters $\tau_{starless} < 0.6$ Myr. The second component is a study of gas density and turbulence. The probability distribution of gas density in the interstellar medium is generally thought to be governed by turbulence, which robustly delivers a lognormal probability distribution for density and velocity. Turbulence requires a driving force on large scales to maintain such a distribution, so I examined both its potential drivers and measures of the distribution. Formaldehyde was used as a densitometer to examine the internal conditions of BGPS clumps. The density analysis yielded information about the distribution of density in active and quiescent clouds. \begin{itemize} \item The density of infalling gas around \uchii regions is comparable to the mean density observed in the interstellar media of starburst galaxies, with typical number densities $10^{4.5}\percc\lesssim n \lesssim10^{5.5}\percc$ \item The comparison of hydrogen and carbon radio recombination line velocities with \formaldehyde gas velocities can be used to determine the evolutionary state of individual \uchii regions \item The density of gas in quiescent GMCs is 1-2 orders of magnitude higher than typically assumed, with $10^4\percc\lesssim n \lesssim10^{4.5}\percc$. This discrepancy indicates either an extreme failure of the spherical cloud assumption, such that the true densities within GMCs are uniformly higher over a smaller volume, or that the density distribution is not governed by normal turbulence. \item Study of the W5 region showed that molecular outflows from young and proto-stars do not drive the turbulence observed in this region. Instead, the turbulence is likely driven along the bubble edges by O-star winds and radiation. \end{itemize} %Therefore, in conclusion, I conclude. \input{solobib} \end{document}